How the actional suffix chain connects Japanese
to Altaic
Martine Robbeets
Robbeets, Martine 2007. How the actional suffix chain connects Japanese to Altaic. Turkic
Languages 11, xx-xx.
In the much debated Japanese-Altaic affiliation question linguists seem to agree on at least
one point: comparative morphology can contribute to unravelling the problem. The present
paper studies verbal morphology. The aim is to relate Japanese derivational and actional
suffixes to Altaic equivalents. The observation made in Johanson (1992: 2002) about the
conservativism of the positions close to the primary stem is taken as a methodological prerequisite. By undoing the lexicalization that took place in the verbs of Old Japanese, an
actional suffix chain is reconstructed. In the body of the article Altaic cognates are proposed for the reconstructed suffixes. The stability of the proposed cognates is assessed in
terms of general cohesion and systematics, parallellism of combinational patterns, equivalence of the basic parts of speech, phonological correspondences and semantic latitude.
Martine Robbeets, Seminar für Orientkunde, Johannes Gutenberg-Universität Mainz,
DE-55099 Mainz, Germany.
1. Introduction
The linguistic origin of the Japanese language is among the most disputed questions
of language history. Attempts have been made to establish a genetic relationship
between Japanese and a vast variety of languages and language families. In fact,
there is no other language in the world which could claim a wider range of attempts
in this matter. Hypotheses have been presented, connecting Japanese to Sumerian,
Indo-European, the Papua languages, Austro-Asiatic, Sino-Tibetan, Tamil, Ainu,
Austronesian, Altaic. In the past decades, interdisciplinary research, joining the
forces of archaeology, genetics, physical anthropology and general linguistics, has
advanced serious support for setting up an Altaic hypothesis of Japanese language
origin. The Altaic hypothesis for Japanese refers to the idea that Japanese is related
to Korean and to the Tungusic, Mongolic and Turkic languages. The cover term
‘Altaic’ can be used in reference to this large group of adjacent languages, stretching
from the Pacific in the East to the Black Sea in the West, because they share a number of properties. Using the term ‘Altaic’ does not necessarily presuppose that the
shared properties are attributable to common ancestorship.
Linguistic literature reflects a wide range of opinions on the Altaic question for
Japanese, ranging from a negative stance over a neutral or agnostic attitude to a posi-
2
Martine Robbeets
tive stance.1 Given the lack of consensus in the field, Robbeets (2005a) presents a
state of the art for the etymological evidence relating Japanese to Korean, Tungusic,
Mongolic and Turkic. The different Altaic etymologies proposed in the scholarly
literature are assessed in terms of phonological and semantic regularity and plausibility. The investigation of the lexical evidence roughs out Japanese as a member of
the Altaic family. However, very few cognate morphemes withstand the sifting process. The morphological evidence is severely underrepresented vis-à-vis the lexical
evidence. And yet, in well-established language families like Indo-European shared
morphology is particularly telling. Safe comparative historical linguistics requires
intercourse between stems and suffixes, it requires lexical evidence to be confirmed
by morphological evidence.
The fact that morphology yields poor results in the Japanese-Altaic case can
partly be attributed to a number of structural features, such as the agglutinative nature
of the languages involved, frequent processes of grammaticalization, the monosyllabicity of the morphemes. But it is also a consequence of our over-emphasis on
lexical research during the last century. Exceptions are Ramstedt 1912, 1952, Poppe
1972, Baskakov 1981, Kormušin 1984, but these studies do not include Japanese
data. Particularly in Ramstedt’s work some of the proposals are outdated in the light
of more recent contributions to the description and reconstruction of the individual
languages and language families.
Comparative studies of Altaic morphology including Japanese are rather limited,
either by the scope of the study or by the number of branches under comparison.
Even in a massive collection of Altaic look-alikes like Starostin, Dybo & Mudrak
(2003), the main focus is on lexical etymologies, and only peripheral attention is paid
to morphological evidence. The most extensive treatment of Japanese morphology in
relation to Altaic can be found in Miller’s work with a study of deverbal verb suffixes (1981), denominal verb suffixes (1982), negation (1971: 245-284, 1985), gerunds (1971: 285-291) and case (1993). Menges treats transitivity pairs (1975: 32-35),
negation (1975: 96-109, 1984: 262-263), gerunds (1975: 110-111) and case (1960,
1975: 111-121, 1984: 245-247). Vovin (1998) provides a sketch of comparative Altaic morphology for Japanese, but in a later article (2001) he restricts the same evidence to an exclusively Korean and Tungusic context. Finally (2005) he claims to
end the Altaic controversy, implying a wholesale rejection of his former contribu1
Doerfer (1963-1975, 1974), Unger (1990), Nichols (1992), Janhunen (1992, 1994), Kiyose
(2002), Shōgaito (2002), Vovin (2005) argue against the hypothesis that Japanese is an
Altaic language. Lewin (1989: 114), Shibatani (1990: 118), Comrie (1990: 856), Lyovin
(1997: 114), Johanson (1999: 2), Trask (2000: 16), Lee & Ramsey (2000: 5) remain
undecided. Ramstedt (1924), Murayama (1975), Miller (1971), Menges (1975), Miller &
Street (1975), Street (1977), Finch (1987), Starostin (1991), Vovin (1994), Kortlandt
(1997), Ramer, Vovin & Sidwell (1997), Sohn (1999: 22), Wang & Ogura (1999), Itabashi
(2001), Starostin (2003) argue in favor of the hypothesis.
How the actional suffix chain connects Japanese to Altaic
3
tions. Finch reconstructs Altaic verb classes (1987) and deals with case (1985, 1999).
Case is by far the most exploited area of Japanese-Altaic comparative morphology,
including an early study by Murayama (1957) and a number of contributions by Itabashi (1988, 1989, 1990, 1991). Binary comparisons of Japanese and Korean morphemes are found in Whitman (1985) and Martin (1968, 1990, 1991a).
Among the articles centered on Japanese morphology in relation to Altaic, many
are restricted to a certain subproblem or to a preliminary exploration of the matter.
Others are partial or small-scale investigations including binary comparisons between Japanese and Korean, eventually adding Tungusic to the initial set-up of the
hypothesis. Genetic divergence can be pictured as the rings formed when a stone is
thrown into the water. Innovations start in the centre of a linguistic area and push the
older forms towards the periphery. Therefore we expect to find traces of conservative
elements in geographically and linguistically remote areas, for instance in Japanese
and Turkic, but not in the other Altaic languages. This observation urges for a global,
holistic approach to the matter.
Although valuable work has been carried out by scholarship in the past, in-depth
and broad-scale research that systematically compares Japanese morphemes to Altaic
is still lacking. Even if the Altaic field is known for its -at times heated- debates,
many scholars seem to agree at least on one point, namely that morphology is a topic
which could contribute to unravelling the Altaic question.
2. Theoretical prerequisites
2.1. Why actionality?
A genetic argument follows a negative argumentation. It works by elimination (Harrison 2003: 215). Arguing for or against a linguistic unity consists in the presentation
of a set of similarities holding between the languages compared. If the similarities are
significantly regular we can, with a certain degree of probability, exclude sheer
chance as their motivation. A genetic argument further consists in the demonstration
that the shared properties are not the result of universal tendencies in linguistic
structuring or language contact. The elimination of universals and borrowings is
based on the tendency of some parts of the language to be more stable and conservative than others.
Morphemes are more resistant to borrowing than lexemes. Unbound morphemes
are more readily copied than bound morphemes. Verbal morphology is more stable
than nominal morphology. The description of borrowability as a relative tendency
leads to the assumption that bound, verbal morphemes provide rather reliable evidence to demonstrate common ancestorship. Johanson (1992, 1999, 2002) has further
refined our notion about borrowability -and hence genetic stability- defining a restricted core of verbal inflectional affixes that is highly telling in matters of genetic
4
Martine Robbeets
relatedness. Johanson (1999: 8) finds that “In the verbal flection, suffixes closest to
the primary stem, markers of actionality and diathesis, seem relatively little susceptible to copying. It would be a strong clue to a common origin if this ‘intimate’ part of
verbal morphology exhibited systematic correspondences of materially and semantically similar morphemes with congruent combinational patterns.”
In his foreword to Johanson (2002), Comrie (2002: xi) highlights that “... in particular the extreme resistance to copying of the positions closest to the verbal stem
might provide a more reliable tool than many of those used in the past to whether
there are indeed shared elements that testify to genetic relatedness, ..., among the
groups of languages that constitute Altaic.” Using Johanson’s findings as a methodological guideline I am confident that actional suffixes provide a reliable starting
point to study verbal morphology relating Japanese to Altaic.
The convincing power of morphological evidence is a matter of competing
forces. A major weakness is the fact that the compared units are monosyllabic, if not
monophonemic. This raises the chance of accidental similarity. Another reservation
is that the number of actionality morphemes in use is low. The formal correspondences will not be recurrent enough to establish phonological correspondences. A
morphological study like the one I intend to undertake is ideally preceded by the
establishment of regular sound correspondences on the basis of lexical data. For this
purpose I refer to Robbeets 2005a. A third point of issue is that lexicalization in progress can obscure the original meaning of the morpheme under inspection. This can
lead to inaccurate semantic reconstruction.
The weaknesses of the evidence are overpowered by the strengths. The occasional monophonemic structure makes it more difficult to pronounce the morpheme
in isolation. The low number of applicable units increases the frequency of use. The
lexicalization in progress leads to mapping intransparency. When the relationship
between form and meaning becomes less clear, the suffix is not easily perceivable as
a distinct unit. These factors make the actional suffixes more resistant to borrowing.
The stability of the evidence is also interrelated with boundness, the internal position,
occasional variant allomorphy and low semantic redundancy. These factors decrease
the perceptibility of the actional suffix and its susceptibility to phonological erosion.
The power of shared morphology is less in the individualism of a single match
than in the determinism of the system of matches as a whole. A convincing argument
for genetic continuity is when several elements known to be quite unsusceptible to
borrowing are retained together. Hence, the strength of the evidence also lies in its
overall cohesion within a paradigm.
2.2. What is actionality?
Actional suffixes derive verbs from lexical stems and modify the basic meaning of
this stem. Morphological derivation and semantic modification characterize actional-
How the actional suffix chain connects Japanese to Altaic
5
ity. German, for instance, is particularly rich in actionality, such as verbal diminutive
(e.g. lächeln ‘laugh a little’ from lachen ‘laugh’), intensive (e.g. verspüren ‘feel consciously’ from spüren ‘feel, sense’), inchoative (e.g. abfahren ‘start the action of
driving’ from fahren ‘drive’), resultative (e.g. aufessen ‘conclude the action of eating’ from essen ‘eat’), iterative (e.g. Heulerei ‘constant bawling’ from heulen ‘bawl,
howl’), momentaneous (e.g. aufschreien ‘cry out, cry loudly and shortly’ from
schreien ‘cry’), durative (e.g. fliegen ‘fly’ vs. durchfliegen ‘fly nonstop’). On the
basis of semantic criteria we can distinguish between three types of actionality, according to (1) the degree of intensity of the event (e.g. low in lächeln vs. high in
verspüren), (2) the frequency of occurrence of the event (e.g. once in abfahren vs.
multiple in Heulerei), (3) the development of the event in the course of time (e.g.
limited in aufschreien vs. delimited in durchfliegen).
Can we refer to the suffix class under inspection as actionality? For deverbal verb
derivation the answer is positive in reference to Steinitz (1981: 41): “… das Vorhandensein einer grammatischen Beziehung zwischen einfachem und abgeleitetem Verb,
die mit der semantischen Beziehung zwischen aktionsartneutraler Grundbedeutung
und zusätzlicher Aktionsartmodifizierung korreliert, [ist] konstitutiv für die Annahme einer Kategorie ‘Aktionsart’ in dem System einer Sprache”. Is it also justified
to include derivational suffixes, suffixes that derive verbs from nouns, adjectives and
adverbs in the present study? In the Japanese verb derivation below, it is observed
that a number of deverbal actional suffixes have counterparts in denominal, deadjectival or deadverbial derivation. They are formally, semantically and combinationally
equivalent. Actional suffixes such as the processive pJ *-na-, the inclinational pJ *ma- and the inchoative pJ *-ka- also function as derivational suffixes. The same
tendency is perceptible in the other Altaic languages. For the present purpose it is
relevant to treat these derivational suffixes under the denominator of actionality,
regardless of the word class of the derivational base.
Actionality can be distinguished from diathesis. Diathetical suffixes do not only
modify the meaning of the preceding segment, they also alter the valency of the base
verb. They can predetermine the syntactic environment of the verb in placing certain
requirements on the surrounding constituents. Diathetical suffixes are suffixes of
voice such as medial, reflexive, reciprocal, cooperative, passive, causative, intransitive, etc. They logically occupy a position following the actional suffixes in the suffix chain.
Actionality contrasts with aspect. Where actionality is the difference between
work ‘spend time and energy’ and work out ‘successfully spend time and energy’,
aspect is the difference between I worked on morphology and I was working on morphology as Lars entered. Actionality derives verbs from any lexical stem, whereas
aspect operates exclusively on verbs. Actionality modifies the base semantically on
word level, whereas aspect gives a different viewpoint on the expression on sentence
level. The semantic modification characteristic for actionality can relate to the tem-
6
Martine Robbeets
poral development of the event, but it can also characterize the degree of intensity or
the frequency of occurrence. Aspect only refers to the internal temporal structure of
the sentence. Actional suffixes logically precede markers of voice and negation,
whereas aspect markers follow them. Actionality is usually expressed by affixes, i.e.
morphological means, as opposed to aspect, which is expressed by verb forms, i.e.
lexical means. Actionality is lexicalization in progress. It gradually converts separate
morphemes into lexical distinctions in verbal meaning. Aspect, on the other hand, is
grammaticalization in progress. It systematically develops a grammatical status for
formerly independent lexemes.
Lexicalization in progress explains why the actionality suffixes are meagerly represented and are lacking productive members in the contemporary Altaic languages,
while they are somewhat better distinguishable in the oldest written stages. If we are
able to undo the lexicalization, the actional suffixes should be reconstructable as
separate morphemes for the individual proto-languages. The reconstruction of the
actional morphemes in the individual proto-languages will be based on internal data
alone. For well-established language families like Indo-European, it is methodologically sound to switch back and forth between internal and external evidence. However, for language families like Altaic, where the actual genetic relationship is still a
matter of debate, the two-way traffic of external and internal evidence becomes a
methodological paradox. Since presupposing an Altaic genetic unity would be methodologically circular, I do not reconstruct actional suffixes in reference to external
comparative evidence.
2.3. Noun, adjective, verb
In Altaic literature we find a misconception relating to the distinction between nouns
and verbs. In my opinion this misunderstanding curbs progress in the comparative
study of derivational and actional suffixes. Miller (1982: 391) contends that “Japanese does not invariably exhibit the same rigid distinction between nominal and verbal roots, stems and suffixes that is seen e.g., in Turkic and Mongolian. In this respect, then, Japanese of all historical periods is significantly closer to Tungusic...”.
This viewpoint elaborates on Menges’ (1978: 289) observation about Tungusic that
“there is a relatively great number of stems which basically are neither exclusively
verbal, nor exclusively nominal, so that they can function in either capacity”. The
idea that there is an interface between nouns and verbs is incorrect. Japanese, Korean
and Tungusic reference grammars clearly distinguish nouns and verbs as distinct
parts of speech, and there is no reason to suppose that the distinction worked in a
different way in earlier periods. Verbs indicate phenomena which take place during
time: activities, processes, states. They are morphologically marked by categories of
voice, aspect, mood, tense and person. Grammatically they are the center of the sentence because of valency. Nouns refer to entities such as animates, body parts, physical objects, and natural phenomena. They take specific inflectional morphology for
How the actional suffix chain connects Japanese to Altaic
7
number, case and possessive and they are characterized by typical derivational patterns. Grammatically, they serve as the heads of noun phrases.
However, studies on individual Altaic languages do not systematically describe
adjectives as a discrete word class. The Japanese adjective is sometimes called a
quality verb (Vovin 2003: 187) because it is like the verb in attaching endings to
mark syntactic function. Adjectives such as J aka- ‘(be) red’, OJ aka- ‘(be) clear,
bright, red’, J taka-, OJ taka- ‘(be) high’, J kata-, OJ kata- ‘(be) hard, tough’ have a
conjunctive form J, OJ aka-ku, taka-ku, kata-ku, a conclusive form J aka-i, taka-i,
kata-i, OJ aka-si, taka-si, kata-si and an attributive form J aka-i, taka-i, kata-i, OJ
aka-ki, taka-ki, kata-ki. The adjective can enter nominal compounds as an uninflected
stem, e.g. J akatonbo ‘red dragonfly’ (< aka- ‘(be) red’ + tonbo ‘dragonfly’), OJ
takayama ‘high mountain’ (< OJ taka- ‘(be) high’ + yama ‘mountain’). A common
view (Unger 1977: 55, Unger & Tomita 1983, Martin 1987: 802, 807) is that the
Japanese adjective inflection is a relatively young phenomenon. This viewpoint does
not exclude the possibility that the quality verbs were originally a subclass of verbs.
Martin (1987: 802) remarks that “The adjective stem is not so free as the noun (you
don’t name things with it), but it is more independent than the verb stem.” However,
quality verbs can be kept clearly distinct from nouns. They are semantically different
because they describe properties of entities whereas nouns refer to entities. They are
also morphosyntactically distinct. A number of quality verbs have nominal counterparts which refer to entities that bear the property or to abstract concepts that express
the property, but they are derived by morphological means, e.g. J take, OJ take2
‘stature; height’ (< *taka- ‘(be) high’ +-(C)i nominalizer). In case of zero-derivation,
when stems of Japanese quality verbs are used as nouns, there is a difference in accent register (e.g.J aka- A ‘(be) red’ and aka 2.5. ‘redness’). This has led to the reconstruction of an original denominal suffix *-m that became abraded, e.g. aka 2.5. <
pJ *aka-m. (Polivanov 1924: 146, Vovin 1994: 250, 2001: 187).
Quality verbs can be distinguished on the basis of their morphosyntactic behavior
from another class of adjectives that does not take inflectional morphology and that
follows different derivational patterns. Adjectives such as J sizuka na, OJ siduka nar‘quiet’ are defined by some Japanese linguists as keiyoudousi ‘adjectival verbs’, by
some western scholars as adjectival nouns (Martin 1991b: 176-77, Kaiser 2001: 5,
Vovin 2003: 93) and by others as na-adjectives (Takeuchi 1999: 81, Backhouse
1984). They are modifiers that require the modifying form na or the adverbial form
ni of the copula in contemporary Japanese. In Old Japanese they are followed by
various forms of the copula nar- ‘be’ or by ni, an adverbial form of the defective
verb n- ‘be’. Some na-adjectives have underived nominal counterparts that refer to
entities that bear the property or to abstract concepts that express the property (e.g. J
heiwa na ‘peaceful’, heiwa ‘peace’). The nominal counterpart can take nominal inflectional morphology, such as for instance the genitive case no in heiwa no sisya
‘messenger of peace’ in comparison to the adnominal form na in heiwa na kuni
8
Martine Robbeets
‘peaceful country’. On the basis of semantics (description of properties vs. reference
to entities), grammatical position (modifier vs. head) morphosyntax (adjectival vs.
nominal morphology) na-adjectives can be distinguished from nouns. In what follows I will refer to quality verbs (e.g. OJ kata- ‘(be) hard’) and adjectives (e.g. OJ
siduka ‘quiet’). Semantically, quality verbs are a subclass of adjectives.
The Korean adjective is inflected and it is considered a subclass of the verb
(Martin 1992: 89). In Korean linguistic terminology, ‘adjective’ is used in reference
to a quality verb, a descriptive intransitive verb (e.g. K pwulk-, MK pulk- ‘be(come)
red’, K kwut-, MK kwut- ‘be(come) hard’, K noph-, MK nwoph- ‘be high’). The stem
of a quality verb cannot stand alone. Some of the descriptive verbs can be traced
back to derived nominal roots (e.g. K mwulk-, MK mulk- ‘be watery, be thin’ < MK
.
mul ‘water’) or to derived verbal roots (e.g. K wusup- ‘be comical’ < K wus‘laugh’). Others are original quality verbs or descriptive extensions from originally
processive intransitive verbs (e.g. K cala-, MK cola- ‘reach, grow; be sufficient, be
enough’).
Among the descriptive verbs, there is a subclass of adjectives such as K kanan
ha- ‘be poor’, phikon ha- ‘be tired’, ttattus-ha-‘be warm’, kkaykkus-ha- ‘be clean’,
kattuk-ha- ‘be full’. Only by combining the auxiliary ha- ‘do, be’ can they be inflected. Their roots are called ‘adjectival nouns’ (Martin 1992: 189, 190; Sohn 1999:
206). In general no case suffix can be inserted between the root and the auxiliary ha‘do, be’. Only a few roots such as K kanan ‘poor; poverty’ or phikon ‘tired; tiredness’ can serve also as free nouns referring to an abstract concept and are separable
from ha- by the particle to ‘also’. The so-called ‘adjectival nouns’ describe properties
of entities such as dimension, value and color. They occupy a modifier position and
they rarely take nominal morphology. Therefore, and for reason of uniformity in the
comparative context, I will use the term ‘adjective’ in reference to forms like K
kanan ‘poor; poverty’ and I will refer to forms like K kwut- ‘be(come) hard’ as quality verbs.
A common conception is that adjectives cannot be distinguished in Tungusic.
Gorelova (2002: 145) finds that “Since their main function, which is the attributive
one, is not manifested morphologically, the nouns of quality cannot be opposed to
other nouns and be defined as adjectives as a part of speech”. However, following the
semantic, distributional and morphosyntactic criteria described in Dixon (1982) and
Johanson (2006), the Tungusic adjectives constitute a distinct word class. Tungusic
adjectives describe properties of entities, such as dimension, value, color. They can
secondarily refer to entities that bear the property or to abstract concepts expressing
the property, e.g. Ma. ajige ‘little, small, younger; the little one’, Ma. den ‘high, tall;
height’, Ev. gugda ‘high, height’, Ev. aya ‘good, well; kindness; the good one’. The
adjectives occur in a modifier position, although they can sporadically function as a
head and take nominal morphology. Adjectives make use of intensifying and deintensifying elements and they can enter comparative constructions. Some adjectives
How the actional suffix chain connects Japanese to Altaic
9
govern case, taking a second argument apart from the subject (e.g. Ud. xoŋto
(+ablative) ‘different (from)’).
A number of Tungusic verbs such as Ma. aka- ‘be sad, grieve’, Ma. bere- ‘be
dumbfounded by fright or anger, be lame’, Ma. fere- ‘become old, become deaf,
become dizzy’, Evk. tila- ‘become exhausted’, Ma. tuli- ‘ be overdue, run over a
deadline, expire’, Evk. obdo- ‘become spoiled (of meat)‘, Even obda-‘become tired,
weaken’, Neg. obolo- ‘become poor’, Ma. obdo- ‘become tasteless’ denote properties
of entities. I will specify them as quality verbs and use the cover term ‘adjective’ for
parts of speech such as Ev. gugda ‘high, height’.2
With the exception of Moghol and Mangghuer, which clearly distinguish adjectives as a separate part of speech, probably due to the impact of foreign influence,
adjectives are generally thought to be undefinable as a distinct word class in the majority of the Mongolic languages. It is commonplace to describe the adjective as a
subgroup of nouns in Mongolic linguistic literature. This viewpoint is, among others,
found in Poppe (1954: 40) “There is no morphological difference between substantives and adjectives; all adjectives occur in only one constant form. All words expressing things can function as adjectives and all words expressing qualities can
function as substantives, e.g. modun ‘tree’ and ‘wooden’, maγu ‘bad’ and ‘evil’” and
in Janhunen (2003:10) “There [in Proto-Mongolic] were two major parts of speech
which may be identified as nouns (nominals) and verbs (verbals), combined with two
separate sets of suffixes, respectively. ... Adjectival words were also basically nominal, though their derivatives could function as verbs, ...” In what follows, however, I
will distinguish between nouns and adjectives on the basis of their semantic, distributional and morphosyntactic behavior. Although they can secondarily refer to entities bearing a property or to abstract concepts denoting the property, adjectives
mainly describe properties of entities, such as e.g. WMo. gün ‘deep; depth’, WMo.
öndür ‘high, tall, height’, WMo. ulaγan, Khal. ulān ‘red; redness, the red one’. In the
noun phrase, adjective-plus-noun constructions are not exclusively, but typically
found. Certain derivational patterns are specific to adjectives.
Mongolic has a separate class of adjectives that grammatically behave like verbs
such as WMo. ayu- ‘be(come) frightened, fear (intr.)’, WMo. bayas- ‘be happy,
content, rejoice (intr.)’, WMo. cad- ‘be(come) saturated, be ripe’, WMo. ice- ‘be/
feel ashamed’, WMo. qata- ‘become hard, dry (intr.)’, WMo. qala- ‘be(come)/ feel
warm’, WMo. soγta- ‘be(come) drunk, intoxicated, be in a rut’. In the comparative
2
Apart from the quality verbs, there are a number of contemporary adjectives that may be of
verbal origin as well. On these adjectives, Benzing (1955a: 1037) remarks: “Bei den von
den Verben abgeleiteten Formen finden wir einige, die auf das Vorhandensein ehemaliger
Zustandsverben hinweisen, wie wir sie z.B. im Koreanischen und im Giljakischen finden:
lam. gel.sa.n ‘(es) ist kalt’ (Aorist, nur prädikativ gebraucht) ~ lam. gel.sī (=udh. gili.hi)
‘kalt’ (< ‘kalt seiend’, Part. aor., nur attributiv verwendet), ... ”
10
Martine Robbeets
approach below, I will distinguish between quality verbs (e.g. WMo. qata- ‘become
hard, dry’) and adjectives (e.g. WMo. öndür ‘high, tall, height’).
Similar to the analysis of the parts of speech in Tungusic and Mongolic, the adjectives in Turkic are often considered a subclass of nouns. Erdal (2004: 143) remarks: “The term [nominals] covers nouns (including proper nouns), adjectives,
pronouns and numerals. We speak of ‘adjectives’ as a special sub-class because there
is an (admittedly fuzzy) semantic distinction between the two classes: Adjectives
tend to denote qualities and are used for referring less frequently than nouns.” Another rejection of the distinction between noun and adjective is found in Grønbech
(1936: 24, 26-27). The borderline between the two word classes, roughed out by
Gabain (1950: 64, 148), is specified by Johanson (2006). He defines the adjective as
a distinct word class with respect to semantic and distributional features, morphosyntactic behavior and certain derivational characteristics. Adjectives primarily describe properties of entities, while their secondary and tertiary semantic function is to
refer to entities that bear the property and to abstract concepts, e.g. OTk. kïzïl ‘red’,
MTk. kïzïl ‘red, intense; a kind of red bird’, OTk. täriŋ ‘deep; depth’, OTk. bädük
‘big, great; greatness’. They typically occupy the modifier position and take specific
adjectival morphology. They are modified by adverbs and may be used as adverbs,
can govern case, are marked for comparative degrees, take (de)intensifying elements
and reduplicate as intensives.
Turkic has a separate class of adjectives that are similar to verbs in their grammatical behavior, e.g. OTk. bädü- ‘be(come) big, great’, OTk. isi- ‘be hot’, OTk. kat‘be hard, firm, tough’, OTk. kïz- ‘be red’, OTk. tumlï- ‘be cold’, OTk. tïnči:‘be(come) putrid, smell foul’, OTk. us- ‘be thirsty’, OTk. yeni- ‘be(come) light’. It
can be remarked that a number of adjectives can ultimately be derived from quality
verbs, e.g. OTk. kïzïl ‘red’ from OTk. kïz- ‘be red’, OTk bädük ‘big, great; greatness’
from OTk. bädü- ‘be(come) big, great’. In what follows I will distinguish between
quality verbs (e.g. OTk. kat- ‘be hard, firm, tough’) and adjectives (e.g. OTk. täriŋ
‘deep; depth’).
The derivational and actional suffixes studied in this article are devices for morphological derivation and semantic modification. A clear-cut definition of the part of
speech to which the derivational base belongs is essential for their reconstruction and
comparison. On the basis of the criteria discussed here, nouns, adjectives and verbs
can be distinguished anywhere in Altaic. Although quality verbs form a subclass of
verbs from a morphosyntactic point of view, semantically they are a subclass of adjectives. The semantic reconstruction of the actional suffixes will be more accurate if
we treat the quality verbs along with the other adjectives.
3. Reconstructing actionality in Japanese
Many Japanese verbs appear to be simplex at first glance, but internal analysis reveals that they are covert complex verbs resulting from adding one or more suffixes
How the actional suffix chain connects Japanese to Altaic
11
to lexical roots. Due to processes of lexicalization, the meaning that an individual
suffix adds to the neutral base is often obscured. This can account for the reason why
analyses of the Japanese verbal inventory such as Martin (1987: 665-800) reconstruct
the suffixes formally and combinationally but leave the semantics open. Unger
(1977: 127-142) provides a material reconstruction and adds a broad semantic tag,
i.e. -ra- spontaneous action, -ya- passive, -na- ‘be the same as’, -ma- seemingness or
attempt to achieve, -ka- punctual or iterative action. The reconstruction of actional
suffixes in this section formally confirms Martin’s and Unger’s previous findings.
The approach differs from the previous ones because it separates the suffixes according to the word class of the base before an attempt is made to pinpoint the
meaning. The resulting semantic description refines Unger’s former reconstruction.
The present analysis accounts for formal, semantic, combinational and derivational
characteristics. Due to the simple phonological structure of the Japanese language,
there are many homophonous forms in the lexicon. The same tendency is expected in
the morphology. Homophony in one stage of the language may have been polysemy
in another. Taking into account derivational and combinational characteristics can
prevent us from lumping together homophonous morphemes.
The four most common formants are found at the end of the lexicalized suffix
chain, in a position following the actional formants. They have to do with valency: pJ
*-ta- causative-passive, pJ *-pa- reciprocal, passive, intensive-iterative, pJ *-saexoactive, pJ *-ra- endoactive and pJ *-(C)i- which reverses the transitivity of the
verb base. Diathesis provides an interesting topic for future comparative studies, but
it will not be treated in this article. Here I will restrict myself to actional suffixes. In
the first subsection the relative suffix order of the derivational and actional suffixes is
analyzed. In the second part, I attempt to undo the lexicalization in order to reconstruct the form, the meaning and the derivational behavior of the suffixes. The verbs
underlying the reconstruction are taken from an exhaustive inventory of 1963 Old
Japanese verbs listed in Unger (1977: 83-126). The verbs are double-checked in
Martin 1987 and Omodaka et al. 1967. The pairs underlying the reconstruction consist of a lexeme of neutral actionality, along with its derived counterpart. The counterpart is preceded by the symbol ( > ). If there is no neutral base attested, I provide
different derivations of the same base.
3.1. The actional suffix chain
Martin (1987: 796) lists the shape-populations of suffix strings lexicalized in 2200
contemporary and historic Japanese verb stems. If an actional suffix appears in a
string, it is usually followed by a diathesis suffix. Strings consisting of two actional
suffixes are rare. With the exception of -n(a)-ka(2)- for which I count 24 instances,
other combinations of two actional suffixes are uncommon: -ra-na-ka(2)- 1, -ra-ka(1)4, -ra-ka(2)- 4, -ra-ma- 1, -na-ma- 2, -n(a)-ka(1)- 3, -ya-ka(1)- 5, -ya-ka(2)- 1, -ma-ka(2)1. The (unless otherwise indicated) contemporary Japanese verbs underlying the
12
Martine Robbeets
given number of suffix strings are the following. I only count instances of strings that
derive distinct roots. The verbs hutagu ‘put a lid on, stop up’ / hutagaru ‘be lidded,
stopped up’/ hutageru ‘put a lid on, stop up’ (<*puta ‘lid’-n(a)-ka(2)-Ö), for instance,
are counted as one instance of the suffix string -na-ka(2)-.
-ra-na-ka(2)- (1): 1. usuragu ‘get thin, pale, abate’
-ra-ka(1)- (4): 1. hiiragu, OJ pipirak- ‘smart with pain’, 2. hagurakasu ‘parry, dodge’, 3.
odoroku ‘get surprised, be horror-struck’, 4. todoroku ‘roar, rumble, throb’
-ra-ka(2)- (4): 1. kawaragu ‘get dry’, 2. midarakasu ‘disorder, disturb’, 3. sabiraku ‘run’,
4. suberakasu ‘let it slip’
-ra-ma- (1): 1. hukuramu ‘swell, bulge’
-na-ma- (2): 1. itonamu ‘perform, conduct, carry on’, 2. ubenamu ‘comply’
-n(a)-ka(1)- (3): 1. korogaru ‘roll, tumble’, 2. sayagu ‘rustle’, 3. soyogu ‘rustle, stir,
sway’
-n(a)-ka(2)- (24): 1. aogu ‘fan’, 2. aragau ‘dispute, argue, wager’, 3. hirogeru ‘spread,
extend, widen’, 4. husagu ‘obstruct, close’, 5. hutagu ‘put a lid on, stop up’, 6. isogu
‘rush, vie to get ahead’, 7. kagu ‘smell’, 8. kasegu ‘work, earn (by work), 9. katugu
‘shoulder, carry on one’s shoulder’, 10. matagu ‘bestride, straddle’, 11. muragaru ‘flock
together’, 12. sirageru ‘polish rice’, 13. sinogu ‘endure, stand, find shelter from’, 14.
sitagau ‘follow, obey’, 15. sogu ‘chop off, diminish’, 16. tairagu ‘get flat, become
peaceful’, 17. togu ‘whet, sharpen, polish’, 18. tugu ‘poor [pour?] (tr.), succeed, be next
(intr.)’, 19. tunagu ‘tie, connect’, 20. tumugu ‘spin, make into yarn’, 21. uragasu ‘delight’, 22. unagu ‘hang around one’s neck’, 23. yabusagaru ‘be stingy’, 24. yurugu
‘shake, waver’
-ya-ka(1)- (5): 1. kagayaku, OJ kakayak- ‘shine, glisten’, 2. sasayaku ‘whisper’, 3.
sosoyaku ‘whisper’, 4. tubuyaku ‘mutter, grumble’, 5. tutuyaku ‘murmur, whisper’
-ya-ka(2)- (1): 1. amayakasu ‘indulge, pamper’
-ma-ka(2)- (1): 1. OJ okumake- ‘anticipate’
The shape of the attested strings makes it possible to motivate the following suffix
order.
base +
1
-ra-
2
-ya-
3
-ka(1)-
4
-ma-
5
-ka(2)-
iconic
inclination
inchoative
-n(a)effort
transformat.
process
This suffix chain slightly differs from the order proposed in Martin (1987: 795). The
first position after the base is occupied by a formant *-ra- that derives actions which
How the actional suffix chain connects Japanese to Altaic
13
are carried out with effort on the base. For reasons explained in section 3.2.1. below,
this suffix is distinguished from the diathesis suffix *-ra- that marks endoactivity.
Contrary to Martin’s analysis, I distinguish between an iconic *-ka(1)- and an inchoative *-ka(2)-. Semantic and derivational criteria for the distinction are found in the
sections 3.2.4. and 3.2.6. below. The inclinational *-ma- occupies a position preceding inchoative * -ka(2)- and causative-passive *-ta-. It is repositioned before * -ka(2)because we find one instance of -ma-ka(2)- in the derivation of OJ okumake- ‘anticipate’ from OJ ok- ‘put’ (compare also OJ okumape- ‘anticipate’), but there are no
unambiguous examples of -ka(2)-ma-. The verb kazikamu ‘get numb, weak, wither’
can at best be derived from a root *kasika-, because the only attested alternant is J
kazikeru, OJ kasike2- ‘get numb, weak, wither’. Contrary to Martin’s analysis the
suffix *-ma- is in a position before *-ta- because we find two instances of -ma-ta-,
i.e. ayamatu ‘err, mistake’, ugomotu, uguromotu ‘bulge up’, while no instances of ta-ma- are found. Working along these lines, the overall chain order confirms a logic
principle in linguistic structuring, namely that actionality is expected to precede diathesis in the chain order.
3.2. Undoing lexicalization
3.2.1. OJ -r- < pJ *-ra1. ‘attempt to achieve the base noun’, 2. ‘execute an action with effort on the base
noun’, 3. ‘make use of the base noun’, ‘sound or behave like the base onomatopoeia’,
‘behave like the base adjective’
denominal
1. OJ ki1pa ‘limit, brink’ > OJ ki1pa-r- ‘come to an end, wear out (intr.)’, OJ kubi1
‘neck’ > J kubi-r- ‘strangle (tr.)’, Miyako (dial.): Uechi kïï, Nakasuji cïï ‘fog’
(Thorpe 1983: 288-89) > OJ ki2-r- ‘fog up, get foggy’, OJ kumo1‘cloud’ > OJ kumo1r- ‘get cloudy, cloud (intr.)’, OJ sima ‘piece of marked-off land, territory, quarters,
island’ > J sima-r- ‘bind, restrict, shut tight; be shut, be tight (tr. / intr.)’, OJ se ‘narrow place, valley, narrows/rapids (in a river)’ > J ser- ‘narrow the gap, press for
quick action’, OJ take2 ‘height, stature’ > J take-r- ‘get excited, show spirit/courage’,
J tuka ‘bundle’ > OJ tuga-r- ‘attach, connect, join on, chain on (tr.)’, OJ mo2to2 ‘root,
origin, base’ > J modo-r- ‘return, revert’, OJ kaki1 ‘fence, hedge’ > OJ kagi1-r- ‘set
limits’, 2. OJ saka ‘incline, slope’ > OJ saga-r- ‘descend, go down, sink, hang
down’, J tuka ‘hilt, handle, haft’> J tuka-ra-s- ‘tire, make one weary, use up (tr.)’, OJ
wata ‘sea’ > OJ wata-r- ‘cross over, span, get transferred (tr./ intr.)’, 3. OJ ipo ‘hut’
> OJ ipo-r- ‘lodge in a hut’, OJ ko2to2 ‘word, speech, statement’ > OJ kata-r- ‘tell’,
OJ te2 ~ ta- ‘hand’ > OJ to1-r- ~ to2-r- ‘take, hold in hand’ (Unger 1977: 111).
The denominal suffix pJ *-ra- is found petrified in many Japanese verbal stems,
but it maintains a certain productivity in contemporary Japanese. Japanese verbal
14
Martine Robbeets
stems are essentially resistant to borrowing. Empirically we observe that very few
simplex verbs were copied from Chinese or other foreign languages as such. If the
need for borrowing a verb does arise, there is a clear preference to borrow a nominalized form such as zyogingu ‘jogging’ and add a native pro-verb covering all purposes such as suru or yaru ‘to do’ instead. There are a small number of foreign loanwords that have supplied a nominal base for a Japanese verb stem derived with J r(a)-. Examples include J demor- ‘demonstrate’ derived from J demo ‘demonstration’
which is a borrowing from English, J sabor- ‘cut class’ derived from the French
donor noun sabotage, J azir- ‘agitate’ which is probably a borrowing from English
agitation, J gebar- ‘engage in strong-arm tactics’, derived from J gebaruto ‘strongarm tactics’, borrowed from German Gewalt, J gyuuzir- ‘take the lead, boss’ derived
from a Chinese borrowing gyuuzi ‘ears of an ox’. More examples can be found in
Martin (1987: 673).
onomatopoetic
*pipi (mimetic for quick, light up and down movement) > OJ pipi-r- ‘flutter up’,
*pipi (mimetic for tingling sensation) > OJ pipi-ra-k- ‘smart with pain, produce a
sharp stinging pain’, J pika-pika ‘sparkling, glittering, shining (mimetic)’ > OJ pi1kar- ‘shine’, *we ~ *wa (emotional exclamation) that occurs in the lamenting OJ we, in
OJ we-warap- ‘laugh out loud’, and in J wa-meku ‘scream, shriek’ > OJ we-ra-k‘laugh with joy’, OJ wa-ra-p- ‘laugh’, *susu (mimetic for slurping sound) > OJ susuro-p- ‘slurp’,3 OJ ne ‘sound, crying, weeping’ > OJ na-r- ‘make a sound, ring’, *no
(mimetic for scolding, cursing) > OJ no-r- ‘scold’, OJ no2-ro2-p- ‘curse’, *kisi (mimetic for high-pitched, unpleasant sounds) > J kisi-r- ‘creak, rasp’, *soso (mimetic
for nervous motion) in OJ soso-mek- ‘fidget, move nervously, rush’,4 OJ sosog- ‘ruffle, move back and forth’ > J soso-r- ‘excite, incite, stimulate, stir up’
deadjectival
quality verbs: OJ aka- ‘(be) clear, bright, red’ > OJ aka-r- ‘brighten, redden’, OJ
naga- ‘ (be) long’ > OJ naga-re- ‘flow’, OJ nagarape2- ‘live on, live long’, OJ usu- ‘
(be) thin’ > OJ usu-re- ‘get thin’, OJ usurag-‘get thin’, OJ kusa- ‘ (be) stinky’ > OJ
kusa-r- ‘rot’, OJ taka- ‘ (be) high’ > OJ taka-r- ‘be high, elevated’, OJ pi1ro2- ‘ (be)
3
4
The assimilation of *-ra- to *-ro- in agreement with the quality of the preceding vowels in
OJ susu-ro-p- ‘slurp’ and in OJ no2-ro2-p- ‘curse’ is probably triggered by mimesis.
J -meku is an auxiliary that derives verbs from iconic expression, such as J hatameku ‘flutter, flap’, J kirameku ‘glitter, sparkle’, J ugomeku ‘wriggle, squirm’, OJ sosomek- ‘fidget,
rush; whisper’, OJ sabame1k- ‘murmur, buzz, clamor’, J sasameku ‘whisper, murmur’, J
sazameku ‘make an uproar’, OJ sosomek- ‘fidget, move nervously, rush’, J sosomeku
‘whisper’, J tutumek-‘murmur’, J wameku ‘scream, shriek’, J zawameku ‘be noisy, rustle’,
etc. The auxiliary indicates that the preceding segment is mimetic.
How the actional suffix chain connects Japanese to Altaic
15
wide, broad, vast’ > OJ pi1ro2-r- ‘be widespread’, OJ omo- ‘heavy’ > MJ omo-r‘become heavy, serious’
On the basis of the following observations, the above suffix pJ *-ra(1)- can be
distinguished from a diathetical suffix pJ *-ra(2)- marking endoactivity.5 The latter
derives for instance OJ kakar- ‘hang (intr.)’ from OJ kak- ‘hang (tr.)’ and OJ okor‘arise, happen (intr.)’ from OJ ok- ‘put (tr.)’. Unger (1977: 140) refers to it as a
marker of “spontaneous action, endo-activity”, Martin (1987: 672) calls it “endoactive (intransitive or passive)”).
The first distinction is that pJ *-ra(1)- derives verbs from nouns, adjectives and
onomatopoeia, while pJ *-ra(2)- is a deverbal verb suffix.
Second, there is a difference in meaning between pJ *-ra(1)- and pJ *-ra(2)-.
Whereas pJ *-ra(1)- modifies the lexical base adding the semantics described above,
pJ *-ra(2)- marks a verb base as endoactive. It not only modifies the preceding segment semantically, but also predetermines its syntactic environment. Placing requirements on the relation of the subject to the predicate, pJ *-ra(2)- affects the
valency of the preceding verb.
Third, pJ *-ra(1)- occupies the first position in the suffix chain, closest to the lexical stem, while pJ *-ra(2)- occurs in more outer ranks. Whereas pJ *-ra(1)- can be
followed by other actional suffixes such as -ra(1)-ka- in e.g. OJ werak- ‘laugh with
joy’, pJ *-ra(2)- is always preceded by them, such as -ka-ra(2)- in e.g. OJ tir- ‘scatter,
get scattered’ > MJ tirakas- ‘scatter (tr.)‘ vs. MJ tirakar- ‘get scattered (intr.)’. The
actional suffix pJ *-ra(1)- can be followed by the exoactive pJ *-sa-, as in OJ tukare‘tire, get weary, get used up’ vs. J tukaras- ‘tire, make one weary, use up’, OJ nar‘make a sound, ring’ vs. OJ naras- ‘sound, ring (tr.)’, OJ ki2r- ‘fog up, get foggy’, OJ
ki2ras- ‘cause to fog, make cloudy’. On the contrary, the endoactive pJ *-ra(2)- is
logically exclusive with the exoactive pJ *-sa-.
This is connected with the fourth distinction, namely that pJ *-ra(1)- is of neutral
transitivity as opposed to pJ *-ra(2)-. We find examples where it derives endoactive
verbs, but also examples where it derives exoactive verbs (e.g. J kubir- ‘strangle
(tr.)’, OJ simar- ‘bind, restrict, shut tight/ be shut, be tight (tr. / intr.)’, OJ sibar‘bind, tie, restrict (tr.)’, OJ watar- ‘cross over, span, get transferred (tr./intr.)’). pJ *ra(2)- only derives endoactive verbs and it commonly alternates with an exoactive
counterpart derived with the suffix pJ *-sa-. Examples of such transitivity pairs include MJ itas-‘bring about, achieve’ and OJ itar- ‘reach, come to’, OJ kudas- ‘take
down, put down’ and OJ kudar- ‘go down, OJ no2ko2s- ‘leave’ and OJ nokor- ‘remain, be left’, OJ no2se- ‘load, carry, let ride’ and OJ no2r- ‘ride, be carried’, OJ
oko2s- ‘raise, arouse’ and OJ oko2r- ‘arise, occur’, OJ sugus- ‘let pass’ and OJ sugur‘pass’, OJ tas- ‘add up’ and OJ tar- ‘suffice’.
5
The subscripts (1) and (2) are used to indicate that we are dealing with two different morphemes.
16
Martine Robbeets
Next, suffixation of pJ *-ra(1)- sporadically triggers voicing of the preceding
voiceless obstruent, while derivations with pJ *-ra(2)- show no secondary voicing in
the preceding segment. Examples involving voicing are J tuka ‘bundle’ vs. OJ tugar‘attach, connect, join on, chain on (tr.)’, OJ mo2to2 ‘root, origin, base’ vs. J modor‘return, revert’, OJ kaki1 ‘fence, hedge’ vs. OJ kagi1r- ‘set limits’, OJ saka ‘incline,
slope’ vs. OJ sagar- ‘descend, go down, sink, hang down’. This voicing phenomenon
is described in Miller (1981: 853). It could be triggered by the original phonological
environment: pJ *-ra(1)- derives from the lateral liquid suffix pA *-la-, while pJ *ra(2)- derives from pA *-ra-.
This leads us to an etymological argument. Whereas pJ *-ra(1)- finds a strong parallel in the etymology given under section 4.1., pJ *-ra(2)- probably goes back to a
middle voice suffix in Altaic. In a number of Korean defective infinitives we find a
petrified marker of middle voice K -ul-, such as in K nwukul-e ‘calm down, loosen
up, get milder, become soft (by itself)’ from K nwuk- ‘be soft, be loose (intr.)’ and K
nelpul-e ‘spread/scatter out widely, become wide (by itself)’ from K nelp- ‘be wide,
broad, spacious (intr.)’. Evenki preserves traces of a Tungusic middle voice suffix in
verb pairs as Evk. lamba- ‘stick to, adhere to (intr.)’ and lamba-rā- ‘hold on, stick
(intr.)’, Evk. ñumu- ‘weaken, grow weak (of arms and legs) (intr.)’ and ñumu-rā- ‘be
bloated, be ailing (of arms and legs) (intr.)’. Written Mongolian has a middle voice
marker WMo. -rA-, deriving e.g. WMo. alda-ra- ‘come loose, come off, be freed
from (intr.)’ from WMo. alda- ‘lose, let go, drop (tr.)’ or WMo. ebde-re- ‘break
down, fall to pieces (intr.)’ from WMo. ebde- ‘break, destroy (tr.)’. A relic of middle
voice is also present in OTk. -(I)r-. It generates verbs like OTk. adïr- ‘separate
(tr./intr.), part from (people)’ from pTk *ad- ‘be(come) different (intr.)’ in OTk. adïn
‘other, another’, Karah. aDna- ‘become different, change’ and OTk. ürper- ‘bristle,
stand on end (of hair)’ from pTk *ürpe- ‘be shaggy’ in OTk. ürpek ‘shaggy, disheveled’.
3.2.2. OJ -y- < pJ *-ya1. ‘reach the state expressed by the base noun’, 2. ‘make use of the base noun’,
‘gradually acquire the property expressed by the adjective base, become the adjective
base’
denominal
1. OJ me2 ‘sprout, bud’ > OJ moye- ‘sprout’, OJ moyas- ‘make it sprout’, OJ tupi1
(M 4508) ‘(at the) end’ > OJ tupiye- ‘be spent, wasted, routed’, OJ tupiyas- ‘spend,
waste’, 2. OJ obi1 ‘belt, girdle’ > OJ obiyakas- ‘intimidate, threaten’, OJ obiye- ‘take
fright at’ (cfr. J obik- ‘gird, inveigle’), OJ ya ‘arrow’ > OJ i- (<*i-ya-) ‘shoot’, OJ
iye- (<*i-ya-(C)i-) ‘get shot’
deadjectival
How the actional suffix chain connects Japanese to Altaic
17
OJ wo(-) ‘little (bound)’ > OJ woyas- ‘weaken (tr.)’, OJ woye- ‘get weakened, be
enfeebled’, OJ nipoye- ‘become red, beautiful, fragrant’(cfr. OJ nipop- ‘get red, shine
beautifully, be fragrant’)
quality verbs: OJ waka- ‘young’ > OJ wakaye- ‘get younger, be rejuvenated’, OJ
ko1- ‘saturated’ > OJ ko1ye- ‘grow fertile, get fat’, J koyas- ‘fertilize, fatten, enrich’,
OJ ama- ‘sweet’ > OJ amaye- ‘seek favor, fawn’
It is inviting to treat the deverbal formant pJ *-ya(2)-, that Unger (1977: 130) describes as a passive, as a distinct suffix. The passive suffix is reflected in verb pairs
such as OJ omop- ‘think, feel’ and OJ omopoye- ‘remember, learn, know’, OJ ki1k‘hear’ and OJ ki1ko2ye- ‘be heard, be audible’, OJ mi1- ‘see’ and OJ mi1ye- ‘be seen,
seem, be visible’, OJ mi1yar- ‘view the distance, overlook, survey’, OJ sakar‘flourish, be in high spirits’ and OJ sakaye- ‘flourish, prosper’, pJ *ta- ‘reach’ in
itaru ‘arrive, reach, attain’, itasu ‘do, cause, bring about’ and OJ taye- ‘come to an
end (intr.)’, OJ tayas- ‘end, put an end to, let come to an end (tr.), pJ *ka- ‘go’ in OJ
ik- ‘go’, OJ kayo1p- ‘ply between, commute, frequent, go regularly’, OJ kare2- ‘get
apart, cease, go away’ and OJ kaye- ‘be parted, get apart’. From the examples it
seems that some verbs denoting bodily and mental activities are formed with this
suffix. Martin (1987: 741) analyzes OJ omop- ‘think, feel’ as a compound of omo-,
perhaps reflecting the quality verb omo- ‘be heavy’, and the diathetical suffix pJ *pa-. If this analysis is correct, the suffix pJ *-ya- follows a diathetical suffix in OJ
omopoye- ‘remember, learn, know’. This distributional order supports the description
of pJ *-ya- as a passive suffix. The label ‘passive’ cannot account for the instances of
denominal and deadjectival derivation above. The suffix pJ *-ya- can be polarized as
exoactive (e.g. OJ tayas- ‘let come to an end (tr.)’ < *ta- ‘reach’ + *-ya- + *-saexoactive) and as endoactive (e.g. OJ mi1yar- ‘view the distance’ < *mi- ‘see’ + *-ya+ *-ra- endoactive), which would place it to the left in the diathetical suffix chain.
Besides the internal evidence to treat pJ *-ya(2)- separately, there is comparative
evidence that can shed more light on the problem. Whereas pJ *-ya(1)- fits into the
etymology proposed in section 4.2., the suffix pJ *-ya(2)- could be related to an Altaic
passive. Mongolic has an obsolete -d- and a productive -da- passive (Street 1957: 65,
Poppe 1954: 62, 1955: 253), deriving for instance WMo. dugul-da- ‘be heard, be
audible’ from dugul- ‘hear’ and WMo. ol-da- ‘be found’ from ol- ‘find’. Some verbs
denoting bodily and mental activities in Turkic are derived with an obsolete medial
suffix OTk -(X)d- (Erdal 1991: 642-644, Gabain 1950: 80). This suffix derives verbs
such as OTk. uya-d- ‘be ashamed’ from uya- ‘put to shame’ and OTk. tod- ‘be full,
satiated’ from to- ‘close, block’. Diathetical suffixes relating Japanese to Altaic form
an interesting topic for future research.
3.2.3. OJ -n- < pJ *-na‘develop or process the concept denoted by the base noun’, ‘develop the property
denoted by the base adjective’, ‘develop the state or action denoted by the base verb’
18
Martine Robbeets
denominal
OJ ata ‘enemy, hostility’ > OJ atanap- ‘harm, injure’, OJ aki1 ‘barter, trade, peddling’ > aki1nap- ‘deal in, sell, trade’, J kasa ‘bulk’ > J kasanar- ‘grow in bulk’, OJ
kasane- ‘pile up, layer (tr.)’, OJ kura ‘position established on a higher level’ (JDB
272), OJ kurawi ‘level, rank’ > OJ kurabe- ‘compare’ (< *kura-n(a)-pa-), OJ mapi1
‘gift, offering’ > OJ mapi1nap- ‘bribe’, J ni ‘burden’ > OJ ninap- ‘shoulder, bear,
carry on one’s shoulder’, OJ oto2 ‘sound’ > OJ oto2nap- ‘make a noise’, OJ pusa
‘bunch’ > OJ pusanar- ‘form a bunch, bunch out (intr.)‘, OJ pusane2- ‘bunch together, make into a bunch’, J tuka ‘bundle’ > OJ tukane- ‘bundle (into one) (tr.)’, OJ
tumi1 ‘crime, sin’ > OJ tuminap- ‘punish’, OJ tura ‘row, line’ > OJ turanar- ‘form a
line, stand in a row’, OJ turane- ‘put in a row, line up, link’, OJ to2mo2 ‘companion’
> OJ to2monap- ‘accompany, go with’, OJ tomonape2- ‘let accompany, take along’,
OJ udu ‘treasure, precious (thing)’ > OJ udunap- ‘prize, value’, OJ ura ‘divination,
foretelling’ > OJ uranap- ‘divine, foretell’
deadjectival
OJ maro2 ‘ round (thing), circle’ > OJ maro2b- ‘tumble’ (< *maro-n(a)-pa-)
quality verbs: OJ ama- ‘sweet’ > OJ amanap- ‘cooperate, be nice/friendly to’, OJ
kata- ‘hard’ > OJ katane- ‘lump/bunch together, summarize’, MJ katanas- ‘summarize’
deverbal
OJ ok- ‘put’, OJ oko2r- ‘arise, happen’ ~ oko2s- ‘raise’ > OJ oko2nap- ‘handle’, OJ
tat- ‘stand, be built’ > OJ tatanapar- ‘get piled up’, OJ tatanaduk- ‘pile up (tr.)’, OJ
tatane- ‘fold up, pile up (tr.)’, OJ uk- ‘float’ > OJ ukab- ‘float (intr.)’, OJ ukabe2- ‘let
float, let rise to the surface (tr.)’
3.2.4. OJ -k- < pJ *-ka(1)‘produce a sound or a sensation like the base onomatopoeia’
onomatopoetic
*kororo (mimetic for animal sounds) > ko2ro2ro2k- ‘bark (in a hoarse voice), neigh,
chirp’, OJ ne ‘sound, crying, weeping’ > OJ nak- ‘cry’, *parara (mimetic for quick
up and down motion) > OJ pararakas- ‘make flutter’, *pipi (mimetic for tingling
sensation) > OJ pi1pi1k- ‘smart, be pungent, give a tingling sensation’, *soso (mimetic for a gurgling sound) > OJ so2so2k- ‘pour a liquid with a gurgling sound’, OJ
sawa-sawa ni ‘noisily, turbulently (mimetic)’ > OJ sawak- ‘cause commotion, disturbance, create a fuss’, OJ to2do2 (mimetic expression for knocking on a door or for
the trampling of horse hoofs) > OJ to2do2ro2k- ‘roar, rumble, throb’, *toyo (mimetic
for reverberation) > OJ toyok- ‘clamor, resound, say loudly’, J ugo-meku ‘wriggle’ >
How the actional suffix chain connects Japanese to Altaic
19
OJ ugok- ‘move (intr.)’, OJ ugokas- ‘move (tr.)’, *uta (mimetic for a loud sound) >
OJ utak- ‘roar’
3.2.5. OJ -m- < pJ *-ma1. ‘consider as the base noun’, 2. ‘attempt to reach the position expressed by the base
noun, tend to acquire dimensional properties similar to the base noun’, ‘give the
impression of the base onomatopoea’, 1. ‘regard as or experience the emotional
property expressed by the base adjective’, 2. ‘tend to acquire the property expressed
by the base adjective’, ‘intend or attempt to achieve the state or to carry out the action expressed by the verb base’
denominal
1. OJ ata ‘enemy, hostility’ > OJ atam- ‘regard as enemy, hate’ (JLTT 678), OJ to2ga
‘blame, offence’ > OJ to2game2- ‘censure, find fault with’, J tuka ‘hilt, handle, haft’>
J tukam- ‘grasp, grip’,
2. OJ api1da ‘interval’ > OJ api1dam- ‘take a break, rest’, OJ ki1pa ‘limit, brink’ > OJ
ki1pam- ‘reach the limit, reach an extreme’, OJ kubo ‘hollow (place), depression’ >
OJ kubom- ‘become hollow, cave in, be dented, have a hollow area’, OJ mo2to2 ‘root,
origin, base’ > OJ moto2me2- ‘seek’, OJ siwa ‘wrinkle’ > OJ siwam- ‘get wrinkled’,
OJ se ‘narrow place, valley, rapids (in a river)’ > OJ semar- ‘get narrow, draw near’,
OJ yo2do2 ‘stagnant place (in a river)’ > OJ yo2do2m-’stagnate, be stagnant’.
onomatopoetic
J koro-koro ‘rolling, over and over’, J kuru-kuru ‘round and round, twirling’ > OJ
karame- ‘twine round, entwine, entangle’, J kurum- ‘wrap in, lap in’, *toyo ~ *doyo
(mimetic for reverberation), doyomeku ‘resound, reverberate > OJ to2yo2m- ~ MJ
doyom- ‘resound’, ugo-meku ‘wriggle, squirm’ > OJ ugumot-, J ugomor-, J
uguromot- ‘(earth) bulge up’, J suu suu ‘smoothly (mimetic)’ > J susum- ‘advance,
go forward’
adjectives
2. OJ arata ‘new’ > OJ aratame2- ‘renew, improve (tr.)’, OJ sidu ‘poor, miserable’,
‘quiet’ > OJ sidum- ‘submerge, sink (intr.), get quiet’
quality verbs: 1. OJ kurusi- ‘painful, bitter’ > OJ kurusime2- ‘suffer (intr.)’, OJ niku‘hateful’ > OJ nikum- ‘hate, dislike, reprove (tr.)’, OJ padukasi- ‘ashamed’ >
padukasime2- ‘shame, disgrace, humiliate (tr.)’, OJ yasu- ‘easy’ > OJ yasum- ‘rest,
sleep (intr.)’, OJ wosi- ‘regrettable; lovable, cute’ > OJ wosim- ‘grudge, regret; prize,
value (tr.)’
2. OJ ita- ‘painful, extreme’ > OJ itam- ‘get hurt, damaged, spoiled’ (JDB 78-79), OJ
kata- ‘hard’ > OJ katame2- ‘make hard, lump together, mass together’, OJ ki1yo1‘pure’ > OJ ki1yo1me2- ‘purify’, OJ niko1- ‘gentle, soft’ > niko1m- ‘soften, get gentle,
20
Martine Robbeets
calm down’, OJ naga- ‘long’ > OJ nagame2- ‘prolong voice, recite, drone’, OJ
pi1ro2- ‘broad’ > OJ pi1ro2me2- ‘spread (tr.)’, OJ puka- ‘deep’ > OJ pukame2- ‘make
deep’, OJ sayake 2- ‘cool, clear, pure’ > sayame2- ‘purify’
The Old Japanese deadjectival verb formant -m- is limited in attestation and it is
no longer productive in Old Japanese, but a derived construction is still widely attested and productive in the early records. In the Man’yōshu we frequently find constructions like tabi1-wo kurusi-mi1 (journey-accusative be painful-gerund) ‘because
the journey is painful’ (M 3674) and pi1to2 me2-wo opo-mi1 (person eye-accusative be
many-gerund) ‘because there are many eyes’ (Man’yōshu 207). The element OJ -mi1
in the construction noun phrase-wo adjective-mi1 is usually called a gerund. Vovin
(2001: 191-192, 198) suggests that the Japanese gerund is cognate to the Korean
coordinative gerund -(u/o)mye and the Manchu gerund -me, Tungusic -mi. However,
his etymology is contradicted by the internal Japanese evidence. The presence of the
accusative case marker wo in the Japanese construction calls for a verb and therefore
argues for an interpretation of the -mi1 gerund as the conjunctive form (on -i1) of a
deadjectival verb on -m- ‘regard as’. Against this internal analysis, Martin (1987:
805) argues that “most of the ..mu verbs are intransitive and the transitive counterparts (..mey- <*-ma-Ci-) have an extra counterpart: …”. In Robbeets (2005a: 9293), I follow Martin’s argumentation. However, verbs such as OJ nikum- ‘hate, dislike, reprove (tr.)’ and OJ wosim- ‘grudge, regret; prize, value (tr.)’ above are transitive and illustrate that the actionality suffix -m- is of neutral transitivity. In the denominal and deverbal derivation -m- can derive intransitives, but also transitive
verbs. The transitivity flipper *-(C)i- in OJ kurusime2- ‘suffer (intr.)’ derives the
intransitive counterpart from an original *kurusim- ‘regard as painful, suffer (tr.)’.
The reconstructed transitive is probably the same deadjectival verb as the underlying
one in the construction tabi1-wo kurusi-mi1. I would like to argue that the source of
the Old Japanese -mi1 gerund is a deadjectival actionality suffix of the type illustrated
above that grammaticalized in its conjunctive form.
deverbal
OJ agar- ‘rise’ > OJ agame2- ‘respect, honor, worship’, OJ aye- ‘fall off, spill, drip
(intr.)’, OJ ayas- ‘drop, spill, shed (tr.)’ > OJ ayamat- ‘err, make a mistake’, OJ
ayamar- ‘err, make a mistake, apologize’, OJ ok- ‘put’ > OJ okumake- ‘anticipate’,
OJ kakus- ‘hide, conceal (tr.)’, OJ kakur- ‘hide (intr.)’ > OJ kakum- ‘surround’, OJ
kitap- ‘drill, train, forge’ > OJ ki1tam- ‘punish, chastise’, OJ tat- ‘stand, be built’ >
OJ tatam- ‘fold up’, OJ nade- ‘pat, stroke’, OJ nadas- ‘deign to pat, stroke’ > OJ
nadame2- ‘soothe, placate’, *nonso- ‘see’ in OJ nozok- ‘peek, peer’ > OJ nozom‘hope for, wish for, look for, look over, view’, OJ puk- ‘cover’ > OJ pukum- ‘comprise, include (intr.)’, OJ pukume- ‘include (tr.)’, OJ pare- ‘swell (intr.)’, OJ paras‘make swell (tr.)’ > OJ param- ‘get pregnant, get filled, get swollen’
How the actional suffix chain connects Japanese to Altaic
21
Besides the deverbal formant for inclination -m- that is lexicalized in a position
close to the primary stem, we find another suffix, the tentative OJ -(a)ma-, that is still
productive in Old and Middle Japanese (Martin 1991b: 605-615, Vovin 2003: 273282). It is a modal suffix that is found in the outer ranks of the verbal suffix chain. A
number of Japanese linguists label this suffix as the presumptive mood. Vovin follows Martin in calling it a tentative. OJ -(a)ma- denotes a wide range of meanings
such as intention, volition, inclination, suggestion, presumption. Form and meaning
being similar, the modal suffix is probably internally related to the actional suffix.
Perhaps OJ -(a)ma- is a composite derivation of the existential auxiliary pJ *a- ‘be,
exist’ and the actional pJ *-ma-, that later acquired a grammatical status.
3.2.6. OJ -k- < pJ *-ka1. ‘put the base noun to use’, 2. ‘come into a new state similar to the base noun’,
‘enter the state expressed by the base adjective’, ‘come into a new state or begin the
action expressed by the base verb’
denominal
1. OJ kadura ‘crown, hair ornament’ > kadurak- ‘use as a crown, as a decoration or
ornament in the hair’, OJ pane ‘feather’ > OJ pag- ‘fletch (an arrow) by fitting a
feather to a bamboo shaft’ (< *pan(a)-ka-), OJ makura ‘pillow’ > OJ makurak- ‘use
as a pillow’, OJ obi1 ‘belt, girdle’ > J obik- ‘gird, inveigle’, OJ te ~ ta- ‘hand’ > OJ
tak- ‘do something with the hands, use one’s hands (e.g. in order to dress hair, row a
boat, pull a net, guide a horse)’, OJ tuna ‘rope’ > OJ tunag- ‘tie up, fasten, secure
with a rope’, OJ ude ‘arm’ > OJ udak- ‘embrace’, OJ wana ‘trap, snare, lasso’ > OJ
wanak- ‘throttle, strangle, choke off’
2. ata ‘enemy, hostility’ > MJ atake- ‘violate, start a tumult’, OJ mi1du ‘water’ >
mi1duk- ‘get soaked (in water)’, OJ se ‘narrow place, valley, narrows/rapids (in a
river)’ > OJ sek- ‘dam, obstruct’, susu ‘soot’ > J susuke- ‘become sooty, soiled’, OJ
se (~ so-) ‘back’ > OJ so2k- ‘get distant, recede’
adjectives
OJ maro2 ‘round (thing)’ > maro2kas- ‘make round, make into a lump (tr.)’, OJ
maro2kar- ‘form a lump (intr.)’, OJ sira(-) ‘white (bound)’ > OJ sirake2- ‘whiten, get
white (intr.)’, OJ sidu ‘poor, miserable’, ‘quiet’ (*<‘low’) > OJ sizuk- ‘sink (intr.)’,
OJ utu(-) ‘empty (bound)’ > OJ utuke2- ‘get empty’
quality verbs: OJ pi1ro2- ‘wide, broad, vast’ > OJ ta-pi1ro2kas- ‘open the hand and
wave’, pi1rak- ‘open (tr.)’, pi1rake2- ‘open (intr.)’, OJ tapasi- ‘reckless, profligate’ >
22
Martine Robbeets
OJ tapake2- ‘fool around, misbehave, engage in adultery’, OJ ara- ‘rough, course’ >
OJ arake2- ‘fall to bits, get scattered’6
deverbal
OJ kudas- ‘take down, put down, lower, defeat’ ~ OJ kudar- ‘go down, descend’> OJ
kudak- ‘shatter, break’, OJ kudake2- ‘be shattered, get broken’, *nonso- ‘see’ in OJ
nozom- ‘hope for, look for, view’ > OJ nozok- ‘peek, peer’, OJ tir- ‘scatter, get scattered, disperse (intr.)’ > MJ tirakas- ‘scatter (tr.)’, MJ tirakar- ‘get scattered (intr.)’,
MJ torobos- ~ toromos- ‘melt, temper (metal)’ > J torokas-, MJ torakas- ‘melt, scatter (tr.)’, MJ torake- ‘melt, dissolve, scatter, get infatuated’, OJ to2b- ‘fly, jump’ (<
*ton(a)-pa-) > OJ tonakar- ‘jump up’, OJ par- ‘open (ground), clear (land) (tr.)’, OJ
pare- ‘get clear, open (intr.)’ > OJ paruk- ‘clear up, open up, get bright, dispel’, OJ
parukas- ‘make clear’, OJ pasir- ‘run, rush (intr.)’ > OJ pasirakas- ‘make run, bring
to gallop (tr.)’, OJ uke2- ‘receive’ > OJ ukakap- ‘peep through, spy, watch, infer,
visit, inquire, hear’, OJ amaye- ‘seek favor, fawn’ > OJ amayakas- ‘indulge, pamper’.
4. Altaic cognates
What can be considered a match when comparing derivational and actional suffixes
between the languages that constitute Altaic? First, there is a formal criterion. The
subsequent consonant and vowel of the reconstructed Japanese proto-morpheme
must correspond regularly with the phonemes of the individual proto-forms. A formal correspondence will be considered regular if it confirms the findings in Robbeets
(2005a: 373-377).
Next, there is a semantic criterion. When the measure of freedom for the compared meanings is too large, the formal correspondence can result from pure chance.
In order to keep the semantic latitude to a minimum, actionality will be compared to
actionality alone. The suffixes under comparison all modify the preceding segment
semantically without altering the valency of the primary stem, as opposed to diathetical suffixes. Semantic latitude within the category of actionality itself will be permitted to the extent that the compared suffixes fall into the same semantic subclass:
intensity, frequency and temporal development.
Third, there is a morphological criterion. The suffixes under comparison must derive verbs. Besides, we expect a certain congruence between the word class of the
bases from which the verb is derived. Distinction is made between nouns, verbs,
onomatopoetic expressions and adjectives.
6
The register involved in the derivation is incompatible: OJ ara- ‘rough, course’ is A, while
OJ arake2- ‘fall to bits, get scattered’ is B. The semantic latitude is also considerable. Only
later did the verb arake- develop the meaning ‘get rough’ in Japanese.
How the actional suffix chain connects Japanese to Altaic
23
Finally, there is a combinational criterion. The relative position that a suffix occupies in the suffix chain in terms of its distance to the primary stem is taken into
account. Aspectual and modal suffixes that occur in the outer ranks of the suffix
chain and often go back to a lexical origin are unconvincing as evidence. The combinational criterion gains relevance within the larger context of comparative verbal
morphology. The actional paradigm must once have fitted into an ordered original
grammar, which is expected to reflect system properties of a real natural language.
For Japanese, Korean, Mongolic and Turkic I use material from the oldest unambigiously written stages: Old Japanese, Middle Korean, Middle Mongolian, Written
Mongolian and Old Turkic. Old Turkic covers the period from the eighth to the
fourteenth century. The examples consist of some old runic data, but mainly of Old
Uighur and peripherally of Karakhanid. The Middle Mongolian forms are taken from
the Secret History of the Mongols, the genealogy and biography of Chingis Khan,
dating back to the thirteenth century. Written Mongolian refers to the literary tradition of Mongolic from the thirteenth century up to present, but it can be considered as
a conservative Mongolic language in itself. Middle Korean mainly refers to Late
Middle Korean. It is the language written down after the invention of the Korean
script (1446), represented in the alphabetic texts of the fifteenth and sixteenth centuries. Before that time no systematic transcription of the Korean language existed.
Early Middle Korean (918-1446) records are all in Chinese characters. They are
phonologically speculative because every sound value must be reconstructed, and
they are only fragmentarily documented. Old Japanese is the language spoken in the
Nara-period (710-794). It is the language of the literary moments of Japanese: the
poetry of the Man’yōshu, the cultural and geographical records of the Fudoki, the
dynastic chronicles of the Nihonshoki. For the historical study of Tungusic languages
it is unfortunate that written records are nonexistent for most of the languages. The
oldest records are written in Jurchen, dating back to the period when a confederacy
of Tungusic tribes ruled over North China under the dynastic name Jin (1115-1234).
However, Jurchen writing has not been completely deciphered yet, and our knowledge about the Jurchen language is fragmentary and speculative. Much more extensive is the literature that is written in Manchu, when the Manchus were in power in
China during the Qing dynasty (1644-1911). However, this language is highly
sinisized. Because it does not always preserve crucial parts of morphology, I make
additional reference to the contemporary Tungusic languages: Evenki, Even, Solon,
Negidal, Nanai, Ulcha, Orok, Oroch, and Udehe. In the following discussion the
references between brackets refer to the secondary source that makes reference to the
morphological data. The semantic analysis of the morphemes is mine. The lexical
stems and their derived counterparts are individually checked in the dictionaries
listed in the bibliography. Occasionally reference is made to primary sources, texts
with examples of the morpheme in question.
24
Martine Robbeets
For the transliteration of linguistic forms, the present paper uses the Yale system
for Japanese and Korean. As far as the vowel distinctions attested for Old Japanese
are concerned, it employs i1 versus i2, e1 versus e2 and o1 versus o2 for the pre- and
postglided variants. The Middle Korean unrounded vowels [ə] and [ɨ] are represented
by o and u respectively, while wo and wu is used for rounded [o] and [u]. A triangle
∆ is used to represent the now obsolete Middle Korean triangle grapheme. The dots
in the Middle Korean words represent the distinctive pitch of the following syllable:
one dot for high, two dots for rising, and unmarked syllables are treated as low. The
transliteration rules for transcribing the individual Tungusic languages follow the
romanization proposed by Gorelova (2002) for Manchu, Nedjalkov (1997) for
Evenki (with the modifications c, j, y, ñ, i for Nedjalkov’s notations ch, d’, j, n’, y
respectively), Nikolaeva (1999) for Udehe (with the modifications j, y for
Nikolaeva’s notations z, j respectively) and a romanization based on Avrorin’s
(1961) Cyrillic transcription of Nanai. The logic that underlies the modifications is a
consistent use of c, j for the palatal fricatives, y for the palatal glide and ñ for the
palatal nasal. The transliteration of the Written Mongolian forms follows Poppe’s
(1954) conventions with the modifications c, j for Poppe’s notations č, ǰ. For Middle
Mongolian Rybatzki (2003) is followed with the exception of š, γ for Rybatzki’s
notations sh and gh. The transcription of the Turkic forms follows Johanson & Csató
1998. For all languages a macron placed over a vowel is used to indicate length. I do
not reconstruct vowel length for the Old Turkic forms.
4.1. pJ *-ra- < pA *-la- effort (denominal, deadjectival, onomatopoetic)
The Japanese suffix analyzed in section 3.2.1. has been compared by Miller (1981:
853) to a different set of Altaic forms, which he derives from a factitive pA *-l-.
Starostin, Dybo & Mudrak (2003: 186-190) reconstruct pA *-l- but leave the semantics open and do not include the Japanese suffix treated here. Ramstedt (1912: 80,
1957: 195-196) and Baskakov (1981: 68) propose the comparison discussed below
for Tungusic, Mongolic and Turkic, not including a Japanese cognate. Schönig
(2003: 416) explains the parallel between the Mongolic and the Turkic suffix as a
case of morphological borrowing. In my opinion it is difficult to attribute the following similarity to borrowing since we are dealing with a denominal verb suffix
very close to the primary stem. Moreover, the shared properties are spread over four
branches, including linguistically and geographically remote groups such as Japanese
and Turkic. The proposed correspondences are global. Phonologically they obey the
lateral liquid correspondence (Robbeets 2005a: 91). The semantic latitude is kept to
an absolute minimum. Except for Tungusic, where the suffix does not operate on
onomatopoeia, the derivational bases are the same: nouns, onomatopoeia and adjectives.
How the actional suffix chain connects Japanese to Altaic
25
Tungusic
pTg *-lĀ- (Benzing 1955a: 1064): Ma. -lA-, Evk. -la-, Even -lĀ-, Ud. -lA-, Na. -lA1. ‘attempt to achieve the base noun’, 2. ‘execute a difficult action on the base noun’,
3. ‘behave like the base noun’, 4. ‘make use of the base noun’, ‘make like or behave
like the adjective base’
denominal
Ma. -lA- (Gorelova 2002, 235): 1. oron ‘vacant post, vacancy’ > oro-lo- ‘fill in, fill a
vacancy’, songko ‘trace, track, footprint’ > songko-lo- ‘follow in the tracks of, imitate’, 3. bata ‘enemy’ > bata-la- ‘be an enemy, oppose’, gucu ‘friend’ > gucu-le- ‘be
friends with, make friends’, sadun ‘relative by marriage’ > sadu-la- ‘form an inlaw
relationship’, ejen ‘ruler, master’ > eje-le- ‘rule, master, occupy by force, establish
control over’, 4. suhe ‘ax’ > suhe-le- ‘split with an ax’, kobto(n) ‘respect, reverence’
> kobto-lo- ‘treat respectfully’
Evk. -la- (Benzing 1955a: 1064, Nedjalkov 1997: 301): 1. mū- ‘water’ > mū-le‘gather water’, dikte ‘berries’ > dikte-le- ‘gather berries’, mō ‘firewood’ > mō-la‘gather firewood’, 4. auun ‘hat’ > au-la- ‘put on a hat’
Ud. -lA- (Nikolaeva 1999: 170-71): 1. sita ‘child’ > sita-la- ‘adopt’, aŋa ‘night shelter’ > aŋa-la- ‘make a night shelter’, iŋme ‘needle’ > iŋme-le- ‘put a thread in a needle’, mamasa ‘wife’ > mamasa-la- ‘marry, take a wife’, anda ‘friend’ > anda-la‘make friends’, 3. jaŋgä ‘boss, leader’> jaŋgä-la- ‘behave like a boss’, 4. kusige
‘knife’ > kusige-le- ‘stab with a knife’, mäusa ‘gun’ > mäusa-la- ‘shoot’, uniŋa
‘spoon’ > uniŋa-la- ‘scoop with a spoon’, santu ‘fist’ > santu-le- ‘hit with one’s fist’,
cob’o ‘handful’ > cob’o-lo- ‘scoop with one’s hand’, aisi ‘gold’ > aisi-le- ‘gild’
Nikolaeva (1999: 13, 171) observes that the suffix Ud. -lA- regularly attaches to
verbal stems borrowed from Russian in order to adjust the Russian verb to the Udihe
derivational system. Examples are Ud. jawoni-la- ‘ring up’ from Russian zvoni- ‘id.’,
Ud. tancewa-la- ‘meaning?’ from Russian tancewa- ‘id.’. This function is reminiscent of the way in which J -ra- adapts foreign loanwords to the Japanese morphosyntactic frame.
Na. -lA- (Benzing 1955a: 1064, Avrorin 1961: 16, 44-45, Menges 1968: 201): 1.
mue ‘water’ > mue-le- ‘get water’, asi ‘wife’ > asi-la- ‘marry, take a wife’, anda
‘friend’ > anda-la- ‘make friends’, 2. usin ‘garden, field’ > usi-le- ‘grow vegetables,
work in the garden’, 3. amin ‘father’ > ami-la- ‘be like a father to somebody, behave
like a father’, enin ‘mother’ > eni-le- ‘be like a mother to somebody, behave like a
mother’ 4. ilga ‘ornament’ > ilga-la- ‘decorate’, epu ‘torture, torment’ > epu-le‘torment, harass’, paloa ‘hammer’ > paloa-la- ‘hit with a hammer’, pao ‘cannon,
gun’ > pao-la- ‘fire a gun’, oma ‘footwear, shoes’ > oma-la- ‘put on one’s shoes’,
apun ‘hat’ > apo-la- ‘put on a hat’
26
Martine Robbeets
deadjectival
Ma. -lA- (Gorelova 2002: 235): genggiyen ‘bright, clear’ > genggiye-le- ‘brighten,
make clear’, beki ‘firm, strong’ > beki-le- ‘make fast, make firm, strengthen’, akdun
‘firm, strong’ > akdu-la- ‘protect, defend’, hahi ‘urgent, hurried’ > hahi-la- ‘act
quickly or urgently, hurry’, untuhun ‘empty; emptiness’ > untuhu-le- ‘be empty’
Na. -lA- (Avrorin 1961: 16, 44-45): padi ‘separate, detached, independent’ > padi-la‘separate, detach (tr.)’, dedu ‘dear, beloved’ > dedu-le- ‘love, be fond of, care for
(tr.)’, masi ‘strong, firm’ > masi-la- ‘be(come) strong’, arkma ‘angry, malicious’ >
arkma-la- ‘become angry with’, manga ‘strong, fierce’ > manga-la- ‘become
stronger’
Mongolic
pMo *-lA- > WMo. -lA- ~ WMo. -nA- after stem final -m-, -ng- (Poppe 1954: 65),
SH MMo. -lA- (Street 1957: 63, 65, 66, Rybatzki 2003: 65)
1. ‘search or hunt for the base noun, carry out a difficult action in order to achieve
the base noun’, 2. ‘behave like the base noun’, 3. ‘make use of the base noun, spend
the span of time denoted by the base noun’,
‘sound like the base onomatopoeia’
‘behave like the base adjective’
denominal
1. WMo. šibaγun ‘bird’ > šibaγu-la- ‘hunt birds’, ang ‘game’ > ang-na- ‘hunt’,
WMo. ger ‘house’ > ger-le- ‘marry, found a house of one’s own’, SH MMo. hoi
‘forest’ > hoi-la- ‘go/ flee into the forest’, 2. SH MMo. aqa ‘elder brother’ > aqa-la‘behave like an elder brother, dominate’, SH MMo. kö’üci ‘one who deals with a son
> kö’üci-le- ‘act like one who deals with a son, treat like a son’, WMo. arsi ‘hermit’
> arsi-la- ‘live as a hermit’,3. WMo. emegel ‘saddle’ > emegel-le- ‘to saddle’, usun
‘water’ > WMo. usu-la- ‘to water’, WMo. altan ‘gold’ > alta-la- ‘gild’, WMo. em
‘medicine’ > em-ne- ‘cure’, SH MMo. kelen ‘tongue, word, speech, statement’ >
kele-le- ‘say, speak’, SH MMo. aqta ‘riding horse’ > aqta-la- ‘ride horseback’, SH
MMo. cag ‘ time’ > cag-la- ‘spend time, deliberate’
onomatopoetic
WMo. qai interjection expressive of grief > qai-la- ‘weep, cry, shed tears’, γoγuu
‘cry of a rooster’ > γoγuγ-la- ‘cackle’
deadjectival
WMo. qurdun ‘quick’ > qurdu-la- ‘rush, be quick’, SH MMo. öter ‘quickly’ > öterle- ‘hasten, do quickly’, soqur ‘blind (adj.)’ > soqu-la- ‘make blind’
How the actional suffix chain connects Japanese to Altaic
27
Turkic
pTk *-lA- > OTk. -lA- (Erdal 1991: 429-455)
1. ‘search or hunt for the base noun, attempt to achieve the base noun, give birth to
the base noun’, 2. ‘execute a difficult action on the base noun’, 3. ‘make use of the
base noun, spend the stretch of time denoted by the base noun’
‘sound like the base onomatopoeia’
‘consider as the base adjective’
denominal
1. av ‘wild game’ > av-la- ‘hunt wild game’, yer ‘place, land’ > yer-lä- ‘to settle at a
place’, iš ‘work, labor’ > iš-lä- ‘work, do or carry out something’, kagan ‘id.’ >
kagan-la- ‘make somebody a kagan’, buzagu ‘calf’ > buzagu-la- ‘calve’, 2. boguz
‘throat’ > boguz-la- ‘cut somebody’s (or an animal’s) throat’, yagï ‘enemy, hostile’ >
yagï-la- ‘engage in hostilities with someone’, 3. adut ‘palm of one’s hand’ > adut-la‘scoop up with the palm of one’s hand’, söz ‘word, speech, statement’ > söz-lä‘speak, say, talk with somebody’, so ‘chain’ > so-la- ‘to chain, fasten with chains’, at
‘horse’ > at-la- ‘ride a horse’, kïš ‘winter’ > kïš-la- ‘spend the winter somewhere, go
into winter quarters’, yïl ‘year’ > yïl-la- ‘spend a year’
onomatopoetic
orï ‘shout, outcry’ > orï-la- ‘shout’, yïgï ‘lament, yammering’ > ïgla- ~ yïgla‘weep’, tigilä- ‘make a certain sound, perhaps a droning one’, kakïla- ‘produce cackling sounds (of birds)’
deadjectival
aduk ‘fault, defect; useless, bad, excessively heavy’ > aduk-la- ‘find strange, to be
astonished’, agïr ‘heavy, important, burdensome’ > agïr-la- ‘honor, respect’, yavïz
‘bad’ > yavïz-la-‘think badly of something’
4.2. pJ *-ya- < pA *-da- transformation (denominal, deadjectival)
The Japanese transformative *-ya- analyzed in section 3.2.2. will be compared with
regard to its denominal and deadjectival derivation only. Miller (1981: 869-870)
follows Hattori (1959: 396-397) with respect to the Japanese reflex and does not
distinguish between the deverbal and the denominal suffix. Ramstedt (1912: 40-43,
1952: 196-197) proposes the comparison discussed below for the Mongolic and
Turkic cognates.
Tungusic
pTg *-dĀ- (Benzing 1955a: 1064): Ma. -dA-, Evk. -dA- ~ -tA-, Even -dĀ-, Ud. -dA- ~
-tA-, Na. -dA-
28
Martine Robbeets
1. ‘reach the state expressed by the base noun’, 2. ‘cover with the base noun’, 3. ‘use
the base noun for one’s advantage, use as a substitute for the base noun, play with the
base noun’
‘gradually acquire the property expressed by the base adjective, become the base
adjective (often of human subjects)’
denominal
Ma. -dA- (Benzing 1955a: 1064, Gorelova 2002: 236-37): 3. ceku ‘swing’ > ceku-de‘swing in a swing’, yobo ‘fun’ > yobo-do- ‘have fun, joke’, faka ‘wooden stick with a
fork at one end’ > faka-da- ‘hit a ball with a wooden stick (play a type of game)’
Evk. -dA- ~ -tA- (Benzing 1955a: 1064, Nedjalkov 1997: 301): 1. jal ‘thought’ > jalda- ‘think’, gerbi ‘name’ > gerbi-te- ‘be named’, daptu ‘issue of a river’ > daptu-ta‘fall into (a river)’ 2. sekte ‘branches’ > sekte-de- ‘cover with branches’, juke ‘ice’ >
juke-te- ‘become covered with ice’, 3. suke ‘ax’ > suke-de- ‘hew, hack’, ju ‘house,
tent’ > ju-ta- ‘live in a tent’, belegē ‘help’ > belegē-te- ‘profit by somebody’s help’,
ilken ‘doll’ > ilken-te- ‘play with a doll’, amin ‘father’ > amin-ta- ‘consider as one’s
father’, enīn ‘Mutter’ > enīn-tē- ‘ consider as one’s mother’
Ud. -dA- ~ -tA- (Nikolaeva 1999: 170-71): 1. akan ‘back’ > aka-ta- ‘turn one’s
back’, 2. ñöŋo ‘glue’ > ñöŋo-do- ‘cover with glue’, 3. mozoli ‘beads’ > mozoli-te‘wear beads’, tiu ‘stick’ > tiu-de- ‘lean on a stick’, sina ‘sack’ > sina-da- ‘carry a
sack’, suala ‘ski’ > suala-da- ‘ski’, uŋtu ‘tambourine’ > uŋtu-de- ‘play the tambourine’
Na. -dA- (Benzing 1955a: 1064; Avrorin 1961: 17-18, 45; Menges 1968a: 201): 2.
alman ‘tin, zinc’ > alma-da- ‘fill cracks with tin’, nute ‘tar’ > nute-de- ‘cover with
tar’, 3. merxe ‘comb with fine teeth’ > merxe-de- ‘comb with a fine comb’, cokor
‘chisel’ > cokor-da- ‘work with a chisel’, ciun ‘flint’ > ciun-de- ‘strike fire’, xiucen
‘elbow’ > xiucen-de- ‘lean one’s elbows’
deadjectival
Although deadjectival derivation with this suffix is frequent in Manchu, it only sporadically occurs in Udehe and Nanai. I was unable to find examples from the other
Tungusic languages.
Ma. -dA- (Benzing 1955a, 1064, Gorelova 2002, 236-37): balama ‘crazy’ > balamada- ‘act crazy’, goho ‘elegant, dandy’ > goho-do- ‘adorn oneself’, hūlhi ‘confused’ >
hūlhi-da- ‘act in a confused manner, be in a daze’, jili ‘anger’ > jili-da- ‘get angry’,
oshon ‘cruel, brutal’ > osho-do- ‘be cruel, be brutal’, bayan ‘rich’ > bayan-da- ‘become rich’
How the actional suffix chain connects Japanese to Altaic
29
Ud. -dA- ~ -tA- (Nikolaeva 1999: 173-174): gē- ‘bad’ > gē-de- ‘feel bad’, pou-pou
‘dark’ > pou-te-gi- ‘get dark’
Na. -dA- (Benzing 1955a: 1064, Avrorin 1961: 17-18, Menges 1968a: 201)
bayan ‘rich’ > bayan-da- ‘become rich’, xele ‘dumb, mute; stammerer’ > xele-de‘mutter, mumble, stammer’
Mongolic
In Mongolic we find two suffixes -dA- and -d- with similar semantics and overlapping derivational behavior. The former suffix frequently, but not exclusively derives
verbs from nominal bases; the latter frequently, but not exclusively derives verbs
from adjectival bases. We find examples where both suffixes attach to the same
nominal base, such as SH MMo. γodoli-d- ~ WMo. γoduli-da- ‘shoot a horn-tipped
arrow (intr.)’. Therefore, it is not unlikely that both suffixes go back to a single suffix, perhaps pMo *-dA- with a sporadic epenthesis of the final vowel.
pMo *-dA- > WMo. -dA- (Poppe 1954: 64), SH MMo. -dA- (Street 1957: 63,
Rybatzki 2003: 65)
1. ‘reach the state expressed by the base noun’, 2. ‘make use of the base noun’
‘gradually acquire the property expressed by the base adjective, become the base
adjective’
denominal
1. WMo. / SH MMo. cisun ‘blood’ > cisu-da- ‘become bloody, be bloodstained,
smear with blood (tr./intr.)’, WMo. idermeg ‘fray, chip, score’ > idermeg-de- ‘become frayed, chipped or scored (intr.)’, WMo. nere ‘name’ > nere-de- ‘be known as,
give a name (tr. / intr.)’, 2. WMo. daγun ‘sound, noise, voice’ > daγu-da- ‘call,
evoke, read aloud, pronounce (tr.)’, WMo./ SH MMo. arγa ‘trick, craft, plan’ > arγada- ‘deceive, outwit, cajole (tr.)’, WMo. γar ‘hand’ > γar-da- ‘do something with
one’s own hands, take in the hand (tr.)’, WMo. buu ‘gun, firearm’ > buu-da- ‘shoot
with a firearm (intr.)’, WMo. ig ‘spindle’ > ig-de- ‘turn the spindle, spin (tr.)’, WMo.
γoduli ‘horn-tipped arrow’ > γoduli-da- ‘shoot a horn-tipped arrow (intr.)’
deadjectival
WMo. kei ‘air, wind (n.); empty, idle, in vain (adj./adv.); hysterical, insane (adj.)’ >
kei-de- ‘be blown away or off by the wind, fly off; become empty; become hysterical
or insane (intr.)’, WMo. γasiγun ‘bitter(ness), sour (adj./ n.)’ > γasiγu-da- ‘grow bitter or rancid, sorrow, mourn (intr.)’
30
Martine Robbeets
pMo *-d-> WMo. -(X)d- (Poppe 1954: 64), SH MMo. -d- (Street 1957: 63, 64;
Rybatzki 2003: 65)
‘make use of the base noun’
‘gradually acquire the property expressed by the base adjective, become the base
adjective’
denominal
SH MMo./ WMo. duran ‘heart, affection, inclination’ > dura-d- ‘mention, quote,
invoke, implore (gods) (tr.)’, SH MMo. γodoli ‘horn-tipped arrow’ > γodoli-d- ‘shoot
a horn-tipped arrow (intr.)’
deajectival
SH MMo. γamtu ‘together’ > γamtu-d- ‘unite, come together (intr.)’, WMo. urtu
‘long (in time and space); length (adj./n.)’ > urtu-d- ‘become (too) long, last, continue (intr.)’ , WMo. sula ‘weak, poor, loose (adj.)’ > sula-d- ‘weaken, become light,
insufficient (intr.)’, WMo. örgen ‘wide, large, broad, vast, width (adj./n.)’ > örge-d‘become (too) wide, expand, be too loose (of clothing) (intr.)’, WMo. / SH MMo.
belen ‘ready, prepared, in readiness (adj./adv.)’ > bele-d- ‘prepare, make ready, become ready (tr./ intr.)’, WMo. cingγa ‘strong, tight, loud’ > cingγa-d- ‘become too
strong, tight, strict (intr.)’, WMo. ariγ ‘pure, clear’ > ariγ-ud- ‘become pure, clear,
clean, holy (intr.)’
Turkic
pTk * -(A)d- > OTk. -(A)d- (Erdal 1991: 485-492)
‘reach the state expressed by the base noun (often of human subjects)’
‘gradually acquire the property expressed by the base adjective, become the base
adjective (often of human subjects)’
denominal
baš ‘head’ > baš-ad- ‘be or become a leader’, og ‘pause, free time’ > og-ad- ‘tarry,
fall behind’ , kut ‘favor of heaven, good fortune’ > kut-ad- ‘become a blessing, enjoy
divine favor and good fortune’
deadjectival
alp ‘tough, resistant, brave’ > alp-ad- ‘be or become a hero’, kïrgïl ‘grey haired’ >
kïrgïl-ad- ‘turn grey haired’, yagï ‘enemy, hostile’ > yagï-d- ‘be or become hostile’,
yogun ‘thick’ > yogun-ad- ‘become thicker’
4.3. pJ *-na- < pA *-na- process (deadjectival, deverbal)
Ramstedt (1912: 62-64) compares the Mongolic and Turkic suffixes below. Later
(1952: 168-169) he proposes a rather speculative internal analysis of two Tungusic
How the actional suffix chain connects Japanese to Altaic
31
suffixes and adds a so-called composite element to the etymology. Poppe (1972: 14041) updates Ramstedt’s proposal by adding the Tungusic deverbal suffix discussed
below. I am unaware of any contributions advancing the Japanese (cf. section 3.2.3.)
and Korean reflexes that are included here.
The proposed etymology is straightforward as far as the deadjectival and deverbal
derivation is concerned. Phonologically, the nasal consonant corresponds regularly,
and the vowel correspondence is satisfactory. The vowel of pMo *-ni- is somewhat
problematic, but it might be morphologically secondary since pMo *-na- is attested
as well. The absence of the final vowel in the Turkic reflex is consistent with the
other Turkic deverbal suffixes compared in the present article. It is probably connected with the observation that verbs with CVC- structure in Turkic had an additional reduced vowel in proto-Turkic which later disappeared. The gradual loss of
final reduced vowels in Old Turkic as demonstrated by Johanson (1979) is accepted
in Turcological literature (Róna-Tas 1998: 72). Semantically, the common denominator is (spontaneous) development, often of natural phenomena. Therefore, I have
tentatively reconstructed pA *-na- as a processive suffix. The term ‘processive’ is
used as an actional distinction in the sense of process versus action. Processive suffixes derive dynamic events that are not caused by an agent and develop spontaneously such as ‘petrify’, ‘bloom’, ‘grow’. They are usually intransitive. Actions, on
the other hand, are caused by an agent. They can occur in the imperative. The proposed etymology spreads over five branches. All branches display deverbal and
deadjectival derivation. The denominal derivation is closely connected in a semantic
sense, but it is only attested in Japanese and Tungusic. In all languages under comparison the nasal suffix occupies the leftmost ranks of the actional suffix chain.
Korean
pK *-no- > MK -.no- processive (Martin 1992: 261, 716, Yu 1973: 335)
‘spontaneously develop and continue the state expressed by the base adjective’
‘develop and continue the action expressed by the base verb’
Although the processive suffix MK -.no- is lexicalized in contemporary Korean,
verbs such as K iss-nun- ‘elapse, stay’ which derives from K iss- ‘exist, be located’,
it are still productive in Middle Korean grammar. Martin (1992: 261) positions the
suffix in the leftmost ranks of the verbal suffix chain. Only occasionally can it be
preceded by suffixes expressing status. Other markers of actionality, politeness,
mood and aspect follow the suffix. The processive MK -.no- is mutually exclusive
with the resultative MK - .kA- and with the retrospective MK - .tA-.
The effective suffix MK -na- is probably not directly related in this context. It is
likely that MK -na- represents a case of internal grammaticalization by incorporation
of the auxiliary MK .na- ‘emerge’. Besides, its occurrence is restricted to a single
verb, MK .wo.na- ‘end up by coming, ultimately come’ from MK .wo- ‘come’ (Mar-
32
Martine Robbeets
tin 1992: 702, 933). Ramstedt (1939: 130) reports other petrified occurrences of the
effective -na-, but apart from MK khuna ho- ‘be big, mighty, powerful’ (Yu 1967:
708) which can be derived from MK .khu- ‘be big’, I was unable to trace the forms.
In the case of MK khuna ho- ‘be big, mighty, powerful’ we are probably dealing with
the adversative -(u)na that appears in the construction X-una X-un ‘that is very X, as
X as can be’, e.g. K khuna khun ‘ever so big, as big as can be’. I have found
Ramstedt’s example kina- ‘extend’, which is derived from kil- ‘be long’, only in the
context of MK kina kin ‘ever so long, as long as can be’.
deadjectival
All examples involve adjective stems that can function both as a descriptive intransitive verb (‘be X’) and as a processive intransitive verb (‘become X’). Martin (1992:
217) lists 14 stems that underlie complete paradigms as both processive and descriptive verbs. When MK -.no- is attached to an adjective stem, it marks the base as a
processive intransitive verb (‘become X’). It turns the adjective into a verb that denotes a dynamic process and that is not caused by an agent.
MK kwut- ‘be(come) hard’ >
MK wuh-.kwa a.lay nung-.hi selu kwut-no-.n-i i-.ta (1586 Sohak 4:53b)7
(top-coordinative bottom be proficient-adverb reciprocally be(come) hard-processive-modifier-postmodifier ‘fact’ be-indicative assertive)
‘the top and the bottom, they both become fairly solid’
MK hoy- ‘be(come) white, light’ >
e.tuw-u.l-ak twolwo .hoy-no-.n-i (1481 Twusi 7:14b)
(‘be(come) dark-prosp. modifier-postmodifier ‘one of two’ again be(come) lightprocessive-modifier-postmodifier ‘fact’)
‘It gets dark and then again it gets light.’
MK .ha- ‘be(come) big, plentiful’>
MK pwul.hwuy ki .phu-n nam.k-on kwoc ¨tywo-.khwo ye-.lum ha-no-.n-i (1445 Yong 2)
(root be(come) deep-modifier tree-focus flower be(come) good-gerund fruit
be(come) big-processive-modifier-postmodifier ‘fact’)
‘The tree with deep roots, its blossoms become good and its fruit becomes plentiful.’
7
Between parentheses reference is made to the late Middle Korean (1443-1592) and modern
Korean (1592-1900) sources where the sentence can be found. The source is preceded by
the date when the work was created or first published. The abbreviations are Yong for
Yongpi echen ka, Wel for Welin sekpo, Twusi for Twusi enhay, Nam for Nammyeng-chen
kyeysong enhay, Sohak for Sohak enhay and Twusi-cwung for Twusi enhay cwung kan.
How the actional suffix chain connects Japanese to Altaic
33
deverbal
Mod. K atok ho-n kono-n pi wo-no-s-ta (1632 Twusi-cwung 12:25b)
(be dim-modifier be fine-modifier rain come down-processive-emotive-indicative
assertive)
‘A dim fine rain sets in.’
MK ¨manh-i tut-.tolwok ¨etwuk .sin-thi a.ni .ho-no-.n-i (1482 Nam 1:36 b)
(be many-adverb hear-projective ‘extent’ the more believe-postmodifier ‘the fact
that’ negative auxiliary-processive-modifier-postmodifier ‘fact’)
‘The more I hear, the less I believe.’
MK .pam-.kwa .nac-.kwa ¨ yen.thyang .ho-no-.n-i-ngi-.ta (1459 Wel 7:59 b)
(night-coordinative day-coordinative lecture auxiliary-processive-modifier-postmodifier ‘fact’-polite-indicative assertive)
‘They keep lecturing away night and day.’
Tungusic
pTg *-nA- (Benzing 1955a: 1064, 1068): Ma. -nA-, Evk. -nA-, Even -na-, Ud. -nA-,
Na. -nA- ‘spontaneously acquire a condition denoted by the base noun, develop natural phenomena denoted by the base noun’
‘spontaneously acquire a property denoted by the base adjective’
‘spontaneously develop the action denoted by the base verb (often of natural phenomena or bodily activities), carry out the verb base all over oneself’
denominal
Ma. -nA- (Ramstedt 1952: 193, Gorelova 2002: 236): abdaha ‘leaf’ > abdaha-na‘leaf, produce leaves’, bongko ‘bud of a flower’ > bongko-no- ‘bud’, eifun ‘pimple,
swelling’ > eifu-ne- ‘develop a swelling’, suihe ‘ear of grain’ > suihe-ne- ‘put forth
ears of grain’, umiyaha ‘worm’ > umiyaha-na- ‘to get worms (of fruit)’, edun ‘wind’
> edu-na- ‘become windy’, asha- ‘wing’ > asha-na- ‘get wings’, talin ‘lightning’ >
tali-no- ‘flash (of lightning)’
Evk. -nA- (Benzing 1955a: 1064): juseren ‘lightning’ > juse-ne- ‘flash (of lightning)’, ima-na- ‘fall (of snow)’
Ud. -nA- (Poppe 1972: 140; Nikolaeva 1999: 172): b’ata ‘boy, fellow’ > b’ata-na‘become grown up (of a boy)’, xatala ‘girl’ > xatala-na- ‘become a girl’, ute ‘rotten
wood’ > ute-ne- ‘molder away, become old’, lusa ‘Russian’ > lusa-na- ‘become
russified’, mō ‘tree’ > mō-no-‘become a tree’, yakpa ‘canyon’ > yakpa-na- ‘appear
(of a canyon)’, edi ‘wind’ > edi-ne- ‘blow (of wind)’, imā ‘snow’ > imā-na- ‘to
snow’, sogdo ‘steam’> sogdo-no- ‘appear (of steam)’
34
Martine Robbeets
Na. -nA- (Benzing 1955a: 1064, Avrorin 1961: 18)
saksa ‘frost pattern’ > saksa-na- ‘develop frost patterns’, juke ‘ice’ > juke-ne- ‘ice
over, become covered with ice’, jolo ‘stone’ > jolo-na- ‘turn to stone, petrify’, sugbin
‘steam’ > sugbi-ne- ‘vaporize, appear (of steam)’, xuren ‘mountain, hill’ > xure-ne‘become mountainous’
deadjectival
Ud. -nA- (Nikolaeva 1999: 173): maŋga ‘strong’ > maŋga-na- ‘become strong’, ede
‘weak’ > ede-ne- ‘become weak’, ñaŋma ‘cold’ > ñaŋma-na- ‘freeze’, sagdi ‘old’ >
sagdi-ne-gi- ‘become old’, xutaligi ‘red’ > xutalihi-ne- ‘become red’, xoligi ‘yellow’
> xoligi-ne- ‘become yellow’
Na. -nA- (Benzing 1955a: 1064, Avrorin 1961: 18): dai ‘big’ > dai-na- ‘be(come)
big’, masi ‘strong, firm’ > masi-na- ‘become strong’, egji ‘many’ > egji-na‘be(come) many’, sagji ‘old’ > sagji-na- ‘become old’, segjen ‘red’ > segje-ne‘be(come) red’, cagjan ‘white’ > cagja-na- ‘be(come) white’
deverbal
Evk. -nA- (Poppe 1972: 140-141): dukte- ‘hit, pond, beat (tr.)’ > dukte-ne- ‘beat (of
the heart)’, kikī- ‘whistle, pipe’ > kiki-ne- ‘howl (of wind)’, *nasa- ‘stretch, wave’ in
nasasō- ‘wave’ and in nasaka- ‘stretch one’s arms sideways’ > nasa-na- ‘wave one’s
arms, flap its wings’, belge- ‘quiver, shudder, startle’ > belge-ne- ‘be agitated, fret
oneself about something’
Ud. -nA- (Poppe 1972: 140, Nikolaeva 1999: 183) is a suffix that only applies to
intransitive verbs and frequently to inchoative intransitives. It denotes that the action
takes place on the whole surface of the subject or that the subject is multiple. Ud.
tiŋme- ‘fall down’ > tiŋme-ne- ‘collapse over the whole surface, fall into pieces’,
bukta-ga- ‘break (inchoative)’ > bukta-ga-na- ‘break all over, break in several
places’, kakta-ga- ‘crack (inchoative)’ > kakta-ga-na- ‘crack all over, crack in several places’
Benzing (1955a: 1064, 1068), Avrorin (1961: 18, 49-50) and Poppe (1972: 140)
treat the deverbal reiterative suffix Na. -nasi- as a compound of *-na- and *-si-. They
relate the former element to the deverbal suffixes discussed above. One should be
aware that the insertion of morphological boundaries on the basis of external comparison makes the evidence more speculative and, to a certain extent, circular. However, in this case we dispose of an internal indication for the boundary because Na. si- is attested as a separate actional suffix deriving continuous, multiple or occasional
actions (Avrorin 1961: 46). Na. omi- ‘drink, smoke’ > omi-nasi- ‘take to the bottle,
How the actional suffix chain connects Japanese to Altaic
35
smoke from time to time’, kalma- ‘crack, burst’ > kalma-nasi- ‘crack all over, chap’,
gele- ‘ask, request, want’ > gele-nesi- ‘beg, cadge’
Mongolic
pMo *-nA-, -ni- > WMo. -nA-, -ni‘spontaneously develop the verb base, carry out the verb base on oneself’
deadjectival (quality verbs)
WMo. meŋde- ‘be or become excited, confused’ > meŋde-ni- ‘be or become excited,
confused’, WMo. togta- ‘become stabilized, fixed’ > togta-ni- ‘come to rest, calm
down’ ,
deverbal
WMo. jagatu-lca- ‘rub against each other’ > jagatu-na- ‘itch’, WMo. joba- ‘suffer,
worry, grieve’ > joba-ni- ‘feel anxiety, be agitated, fret oneself about something’,
toqu- ‘put one thing over another, put on or over’ > toqu-ni- ‘settle down, fall in line,
become quiet’, WMo. sibe-r ‘whisper’, sib-si- ‘speak in a low voice, whisper’ > sibene- ‘whisper’, WMo. bada-na- ‘babble, grumble’
As explained in the following section (4.4.), I think that the analysis (Poppe 1972:
141, Tekin 1982: 507) of the onomatopoetic suffix WMo. -ginA- as a compound of
iconic pMo *-gi- and reflexive pMo *-nA- is incorrect.
Turkic
pTk *-n-> OTk. -(X)n- (Erdal 1991: 584-639)
‘1. spontaneously develop the verb base, 2. carry out the verb base for one’s own
benefit, 3. carry out the verb base on oneself’
Erdal (1991: 634-637) illustrates cases of -(X)n- derivates that take a direct object. In this way he demonstrates that although the derived verbs semantically display
medial-reflexive behavior, they are not always medial-reflexive in a syntactic sense.
Semantic modification without impact on the valency of the verb can serve as an
indication that we are dealing with an original actional suffix instead of a diathetical
marker.
deadjectival (quality verbs)
ar- ‘be tired, exhausted, weak’ > ar-ïn- ‘tire (intr.) ’, arï- ‘be(come) clean, pure’ >
arï-n- ‘purify oneself, be pure’, isi- ‘be hot’ > isi-n- ‘have warm feelings towards
someone’, kat- ‘be(come) hard’ > kat-ïn- ‘become hard, tough’
deverbal
1. bar- ‘go’ > bar-ïn- ‘flow off from a person’s body’, to- ‘close, block’ > to-n- ‘be
closed, be blocked’, bašla- ‘begin, lead (tr.)’ > bašla-n- ‘begin (intr.)’, bärt- ‘wound
36
Martine Robbeets
or break without inflicting visible injury’ > bärt-in- ‘get injured’, 2. bak- ‘look at’ >
bak-ïn- ‘look for one’s own benefit, look around’, kör- ‘see’ > kör-ün- ‘become visible, be seen, see for oneself’, 3. alk- ‘destroy, use up, finish’ > alk-ïn- ‘consume or
exhaust oneself’, ač- ‘open (tr.)’ > ač-ïn- ‘open one’s clothes, baring the bosom,
disclose one’s sins’, yu- ‘wash (tr.)’ > yu-n- ‘wash (intr.), wash oneself’
4.4. pJ *-ka- < pA *-ki- iconic (deadverbial)
Ramstedt (1912: 36-37) compares the Mongolic and Turkic iconic suffixes and devotes an article (1951) to the topic. Erdal (1991: 468) argues in favor of the connection proposed by Ramstedt. Tekin (1982) updates Ramstedt’s work and adds Tungusic cognates. Miller (1982: 401) advances the Japanese cognate along with a Korean reflex different from the one proposed here.
Iconic pA *-ki-has a complete etymology, stretching over five branches. Except
for Japanese, where all actional suffixes have a final vowel -a-, the correspondences
are regular for consonant and vowel. The meanings and derivational bases are identical. We cannot exclude that the ultimate origin of this suffix is a common Altaic
auxiliary *ki- ‘do, make’. The root may be preserved in WMo./ SH MMo. ki- ‘do,
make, produce’, in OTk kïl- ‘do, make’ and in a number of causative formations
throughout Altaic. The etymology of *-ki(-)rA-, however, is more problematic. Erdal
(1991: 465, 468) contends that “they [OTk -kI- and -kIr-] must have been related in
proto-Turkic or in proto-Altaic (if such a language ever existed).” and that “This
analysis [OTk -kIr- as a composite] is not in contradiction with a connection with
Mo. +kirA-, as that could also be a composite suffix.” Since the so-called composite
iconic suffixes are only attested in Turkic and Mongolic, it is not unlikely that we are
dealing with code-copying.
Korean
K -i- < pK *-(k)i ‘produce a sound or a sensation like the base onomatopoeia’
Korean derives sound symbolic verbs from onomatopoeia (= X) using the constructions X-ha- ‘do X’, X-keli- ‘do X repeatedly or continuously, sound like X’ and,
less frequently, X-i- ‘do X repeatedly or continuously, sound like X’. Since the large
majority of the onomatopoeia followed by K -i- ‘do repeatedly, sound like’ end in a
velar stop -k, it is not unlikely that K -i- goes back to pK *-ki-. We find only few
cases where K -i- ‘do repeatedly, sound like’ occurs after a final liquid -l or velar
nasal -ng. Ramstedt (1939: 140) reports to have found K -(k)i- following onomatopoetic expressions, but I am unable to trace the forms with -ki-. All make use of -iinstead. Martin (1992: 588) is tempted to regard K -i- as a reduction from K (kel)i-.
But given the relative infrequency of the former vis-à-vis the latter and given the
large scale of the proposed reduction, I tend to consider K -i- (< pK *-ki) as a more
conservative element. Miller (1982: 401) does not consider K -i- and proposes K
keli- as a cognate. In light of the comparative evidence (WMo. -kirA- < ? pMo *ki-
How the actional suffix chain connects Japanese to Altaic
37
rA-, OTk. -kIr- < ? pTk *-kI- + *-(I)r-) I cannot exclude that K keli- is a further derivation of pK *ki with *-le- (verba media, cf. section 3.2.1.), but this is speculative,
for it requires metathesis of the Korean vowels.
onomatopoetic
K wumcik wumcik ‘budging, stirring, moving’ > wumcik-i- ‘move, stir, put in motion’8, K kkancak kkancak ‘being persistent’> kkancak-i-/ kkancak-keli- ‘stick to,
adhere to, cling to, be persistent’, K kutek, kkutek, kkuttek ‘nodding, bobbing, making
a slight movement’ > K kutek-i- / kutek-keli- ‘nod’, MK kuteki- ‘nod (one’s head)’, K
tulmek ‘shaking’ > tulmek-i- / tulmek-keli- ‘shake’, K (s)swuktek (s)swuktek ‘in whispers, under one’s breath’ > K (s)swuktek-i- / K (s)swuktek-keli- ‘whisper’, K
(s)swuktel (s)swuktel ‘in whispers, under one’s breath’> K (s)swuktel-i- / K
(s)swuktel-keli- ‘whisper’, tallang tallang ‘frivolously, restlessly’ > tallang-i- /
tallang-keli- ‘act frivolously, be restless, be always on the move’
Tungusic
pTg *-ki- is lexicalized in onomatopoetic verbs: Ma. -ki- ~ -gi-, Evk. -ki- ~ -gi-, Even
-kï- ~ -gï-, Na. -ki- ~ -gi- ‘produce a sound or a sensation like the base onomatopoeia’
onomatopoetic
Ma -ki- ~ -gi-: jor ‘sound of many humans or of screaming animals’ > jorgi- ‘chirp,
twitter, hum’, tur ‘sound of a horse clearing its nose’ > turgi- ‘clear the nose (of
horses), snort’, holor ‘sound of a bell’ > hōrgi- ‘ring’, cargi- ‘explode’, kemki‘gnaw, bite (of dogs, geese)’
With the surface exception of Ma. carki- ‘rattle together (as belt pendants), create
a dissonance, tinkle’, which in reality is a borrowing from WMo. cargi- ‘rattle, make
a harsh sound, speak harshly’ (Rozycki 1994: 45), the suffix usually undergoes lenition to -gi- after the liquid r.
Evk. -ki- ~ -gi-: simki- ‘cough’, uŋki- ‘cry, weep’, ñeki- ‘gnaw, crunch’, hēlki- ‘flash
(of lightning)’, kikī- ‘whistle, pipe’, burgī- ‘flow with a splashing noise’, cirgi‘chirp, twitter’, kiŋgī- ‘sound, ring’, kergi- ‘snort’, sirgi- ‘make a clattering noise,
creak, screak’, dergi- ‘shiver, tremble’, sargi-‘splatter (of water, rain)’
Even -kï- ~ -gï-: hiēmkï- ‘cough’, hargï- 'snort, pant', jōŋï- ‘resound, echo; imitate’
8
In Korean **wumcik keli-/ ha- is not attested, we find wumcil keli-/ ha- ‘moving timidly’
from wumcil, wumccil ‘moving timidly’ instead. This may serve as an indication that the
velar in K wumcik-i- originally belonged to the suffix (? < pK *wumci(l)-ki-).
38
Martine Robbeets
Na. -ki- ~ -gi-: siŋbi- ~ simki- ‘cough’, nirgi- ‘thunder’, xurgi- ‘make noise’, dergi‘shiver, tremble’, fulgi- ‘blow’
Mongolic
pMo *-ki- > WMo. -ki- (~ -gi- following a vowel or -r-)
pMo *-ki- + *-rA- medial ? > WMo. -kirA- (~ -girA- ~ -ginA- following -ng-)
‘produce a sound like the base onomatopoeia’
WMo. -ki- ~ -gi-: cuu ‘sound, noise, echo, rumor’ > cuu-gi- ‘make noise, shout (of
many people), quarrel’, WMo. ša ‘sound of a downpour’ > ša-gi- ‘pour, rain heavily’, car ‘sound of voice, cry, clamour’ > car-gi- ‘make a harsh sound, rattle, speak
harshly’, bar-gi- ‘produce a hoarse sound, speak with a hoarse voice’, tur-gi- ‘snort
(of horses)’, cis-ki- ‘chirrup, chirp, twitter, tweet’, kürd ‘sudden explosive noise’ >
kürd-ki- ‘make noise, shout, talk nonsense’ ~ kür-ki- ‘talk nonsense, chatter indiscreetly’, tüs ‘sound expressing sudden blow or banging noise’ > tüs-ki- ‘make a
crashing sound’, šuu-gi- ‘whistle, rustle, sing, howl (of wind)’
WMo. -kirA-: qas-kira- ~ qaš-kira- ‘shout, scream, yell, howl’, bar-kira- ‘roar, bellow, cry, yell’, ar-kira- ‘growl, snarl’, kür-kire- ‘grow, grunt, snarl, roar (as a waterfall)’, or-kira- ‘roar, bawl, growl, whistle (as an arrow)’
WMo. -girA- ~ -ginA-: qang-gina- ‘sound, ring, resound, whine, creak’ ~ qang-gira‘rattle’, qong-gina- ‘ring, sound resonantly’ ~ qong-giras- ‘ring’, küng-gine- ‘make a
hollow sound’ ~ küng-gires- ‘sound, make noise, neigh’, gang-gina- ‘yelp, give forth
piercing cries, cackle, gaggle (of geese)’, cing-gine- ‘ring, resound’, yang-gina- ‘emit
a sharp, high-pitched sound, ache’, šuugina- ‘whistle, make noise (of wind, waves)’
Tekin (1982: 507) analyzes WMo. -ginA- as a compound of iconic pMo *-gi- and
the reflexive pMo *-nA-, discussed in section 4.3. However, in the large majority of
the examples WMo. -ginA- appears after sound symbolic expressions with a final
velar nasal. Therefore, the alternant can better be explained by nasal assimilation.
This explanation cannot account for WMo. šuugina- ‘whistle, make noise (of wind,
waves)’.
Turkic
pTk *-kI-> OTk. -kIpTk *-kI- + *-(I)r- medial ? > OTk. -kIr‘produce a sound or a sensation like the base onomatopoeia’
How the actional suffix chain connects Japanese to Altaic
39
OTk. -kI- (Erdal 1991: 468, Tekin 1982: 508): OTk. bïrkïr- ‘snort’, OTk. bïrkïg
‘snort (of a horse)’,9 OTk. okï- ‘call, call out loud, recite’, OTk. okï- ‘vomit’, OTk.
tok tok ‘knocking sound’ (onomat.) > OTk. tokï- ‘hit, knock, beat, weave’
OTk. -kIr- (Erdal 1991: 466-467; Tekin 1982: 509-510): ay ‘oh, hi’(exclamation) >
ay-kïr- ‘shout out loud’, (Tk. bar bar ‘bellowing sound’ >) ba-kïr- ‘bellow’, kï: ‘hi!’
(exclamation) > kï:-kïr- ‘shout’, sïkkïr- ~ sïkïr- ‘whistle’, alakïr- ‘shout’, bïrkïr‘snort’, üškir- ‘whistle, hiss’.
4.5. pJ *-ma- < pA *-ma- intention/inclination (denominal, deadjectival, deverbal)
Although I am unaware of any etymologies proposed for pJ *-ma- analyzed as a
suffix denoting inclination in section 3.2.5., Martin (1991a: 285), Unger (2000: 664),
Vovin (2001: 194) propose various Korean cognates for the modal suffix OJ -(a)mathat is probably internally related to pJ *-ma-. Vovin further adds the Evenki desiderative suffix -mu-. Miller (1985: 68-69) compares a derived complex suffix OJ maku (< *-ma- + *-ku- nominalization) to the Turkic infinitives in -mAk. Another
derived suffix, the subjunctive and desiderative marker -amasi (< *-ma- + *(po)si‘be desired’), is compared to various Tungusic suffixes by Miller (1985: 61) and
Vovin (2001: 194-195). I do not support these proposals because they compare forms
that are morphologically complex in Japanese to external look-alikes. Ramstedt
(1952: 181-182) explains the Tungusic forms as independent verbal forms that underwent grammaticalization.
The phonological and semantic correspondences of the etymologies suggested
here are satisfactory. However, the derivational bases do not completely overlap.
Mongolic lacks denominal derivation and in Tungusic and Mongolic few examples
are found of deadjectival derivation. The deverbal suffixes in Korean and Tungusic
are distinct from the denominal and deadjectival suffixes. They have a different
vowel. The denominal and deadjectival suffixes are immediately attached to the
primary lexical stem, while the deverbal suffixes occur in a relative outer position in
the chain. The latter are modal suffixes that are still productive in grammar to denote
desiderative or intentional mood. Although similar in form and meaning, their outer
position makes them weaker cognate candidates for the Japanese actional suffix.
Korean
pK *-m(u/o)- > K, MK -m- (after -l-), K, MK Ø (elsewhere)
9
OTk -g derives deverbal nouns, as e.g. in OTk. bilig ‘mental process’ from bil- ‘know’ and
in OTk. bag ‘bale, bundle’ from OTk. ba- ‘bind, tie, fasten’ (Erdal 1991: 182). The derivation of pTk *bïr-kï- ‘snort’ from *bïr ‘sound made by a horse’ is further supported by
OTk. bïr-kïr- ‘snort’.
40
Martine Robbeets
‘1. attempt to cover the space denoted by the base noun, 2. make use of the base
noun’
‘give the impression of the state denoted by the base adjective’
pK *-ma- > K, MK -ma‘intend to carry out the action denoted by the base verb’
The lexicalized suffix -m- has left a segmental trace following liquid phonemes (l) only. The derived verbs on ¨-lm- all have a rising tone in Middle Korean or length
in Korean, suggesting a disyllabic origin. For nouns ending in nasals (-n/-m) the
same semantic modification is obtained by zero verb derivation. This becomes clear
from the example with -m- derivation, K pāl ‘the span of two arms’ vs. K pālm‘measure in arm spans’ as compared to the example with zero derivation, K pyēm
‘span of a hand’ > K pyēm-‘measure in hand spans’. Although the verbal derivation
of nouns ending in -n, -m is zero segmentally, it does leave a trace suprasegmentally.
In contrast to the base nouns, the derived verbs have a rising tone (¨) in Middle Korean that corresponds to length in contemporary Korean. The parasegmentals suggest
that the derived verbs are disyllabic in origin. It can also be remarked that we find
only three simple verb stems ending on -n in Middle Korean: MK ¨sin-/ si.n(u/o)‘wear (shoes)’, MK ¨an-/a.n(o)- ‘embrace’ and MK ¨ten-/ te.nu- ‘wager, bet’. Two
can be derived from noun bases as in the examples below. These are the verbs of
which Martin (1996: 5) invitingly asks: “Why are there so few ..n-stems to begin
with? Were they derived from nouns?”
denominal
1. K pāl ‘the span of two arms, unit of length equal to the span of two arms, 2 yards’
> K pālm-, MK palm- ‘measure off the length in double-arm spans (2 yards)’, K
pyēm ‘span, span of a hand’ > K pyēm-‘measure by the span, span off with one’s
hand’, K phum, MK .phum ‘width of a coat, bosom, space between the chest and
clothes’ > K phum-, MK .phum- ‘carry in the bosom, embrace, harbor’,10 K an, MK
.
anh ‘interior, inside’ (-h place suffix) > K ān-, MK ¨an-‘hold in one’s arms, embrace’, 2. K sal ‘frame, spoke, teeth (of a comb, etc.), fish spear’ > K sālm- ‘harrow
(the soil), rake (the soil)’, K saym, MK soym ‘source, well, spring’ > MK soym-
10
MK .phum ‘bosom’ may derive as a deverbal noun on -m from MK .phu- ‘bloom, spread,
extend’ (Martin 1996: 25, 47). Martin (1996: 47) remarks that the internal derivation of
MK .phum- ‘embrace’ is ”at the expense of at least part of the comparison with pJ
*pukuma- L ‘hold in the mouth, harbor, comprise, contain’”. But, if Japanese hukumu
‘harbor, contain’ can be derived as pJ *puku-ma- in reference to the analysis in section
3.2.5., the comparison with MK .phum- ‘carry in the bosom, embrace, harbor’ (< *pukum(u)-) remains possible.
How the actional suffix chain connects Japanese to Altaic
41
‘spring up, spurt up’,11 K sin, MK .sin ‘footwear, (Korean style) shoes’ > MK ¨sin‘wear (shoes), use as footwear’
deadjectival
K kwul ‘hole, cave, cavity, empty’, K kwulh- = kolh-, MK kwolh- ‘be empty’ > K
kwulm- ‘starve, skip a meal, go hungry’, MK ¨kwulm- ‘hunger, go without food,
become empty’, K kolum ‘pus, purulent matter’ (K -m deverbal noun from *kwol‘be purulent, be rotten’), K kolh- ‘go bad, rot, spoil, get stale’, MK ¨kwolq- ‘fester
(intr.)’12 > K kolm- ‘form pus, fester’, MK kwolm- ‘form pus, fester (intr.)’, K telep~ talap-, MK ¨telep- ~ ¨talap-, ¨teleW- ~ ¨talaW- ‘be muddy, dirty’ (< *tele- ‘be
dirty’ + pK *- . po- adjective suffix13) > MK ¨telm - ‘become dirty, get stained, get
dyed’
deverbal
Martin (1991a: 285, 1992: 248) and Vovin (2001: 194) propose an etymology for the
modal suffix OJ -(a)ma- that is probably internally related to pJ *-ma-. They relate
the Japanese tentative to the Korean modal suffix K/ MK -ma ‘will, be willing to,
intend to, promise’. The suffix is still productive in Korean and denotes intentions or
assumptions. As a modal suffix it occurs relatively to the right in the suffix chain. Its
outer position weakens the comparison with the Japanese actional suffix.
K Nayil ka-ma.
(Tomorrow go-intentional)
I will come tomorrow
K Ku kes-un kutay-lo ha-y cwu-ma.
(That thing-focus that way-manner do-infinitive give-intentional)
‘I’ll do it just as you wish.’
Tungusic
pTg * -mĀ- (Benzing 1955a: 1064): Evk. -mA-, Even -mĀ- , Ud. -mA-, Na. -masi11
12
13
MK ¨soym ‘source, well, spring’ may derive as a deverbal noun on -m from MK . soy‘leak’.
MK q is a morphophonemic symbol, used by the Yale Romanization to represent all instances of noninitial reinforcement. In Middle Korean q represents a glottal stop that remains unpronounced when initial, but represents reinforcement when final.
MK -W- ~ -p- < pK *- .po- is a suffix that derives depictive adjectives from adjective stems
such as in MK/K kwut- ‘be hard, be firm (in belief)’ > K kwuteW- ‘be gullible, quick to
believe’, MK te .i- ‘warm up’ > MK ¨teW- ‘be warm, hot’. If MK ¨telep- ~ ¨talap-, ¨teleW~ ¨talaW- ‘be muddy, dirty’ can be derived from the adjective base *tele- ‘be muddy,
dirty’ followed by this suffix, the base is perhaps cognate with J doro ‘mud, mire, dirt’.
42
Martine Robbeets
‘attempt to achieve the base noun, hunt the base noun, incline to the position, state or
being denoted by the base noun’
pTg * -mī- (< *-mā- + -gi- causative) (Benzing 1955a: 1065)
‘make, achieve the base noun, attach to the base noun, hunt the base noun’
‘turn into / reach the adjective base’
pTg *-ma-, *-mu- (Benzing 1955a: 1069): Evk. -mu-, Even -m-, -mš-, Neg. -mu-, Ud.
-mūi-, Na. -mosi- ~ -musi‘intend, want, be inclined to carry out the action denoted by the verb base’
denominal
Ma. -mi- (Gorelova 2002: 236): doko ‘lining of a garment’ > doko-mi- ‘line (a garment)’, tohon ‘button’ > toho-mi- ‘button (up)’, ture ‘leg of a boot’ > ture-mi- ‘attach
the leg of a boot’
Evk. -mA- (Benzing 1955a: 1064-65, Nedjalkov 1997: 300): taman ‘cost, price’ >
taman-ma- ‘pay’, uluki ‘squirrel’ > ulu-me- ‘hunt squirrel’, ollo ‘fish’ > ollo-mo- ‘to
fish’, mōti ‘elk’ > mōti-ma- ‘hunt elk’
Evk. -mi- (Benzing 1955a: 1065, Nedjalkov 1997: 301): ollo ‘fish’ > ollo-mi- ‘to
fish’, beyun ‘wild reindeer’ > beyu-mi- ‘hunt wild reindeer’, here ‘lower part, bottom, sole’ > her-mi- ‘sole, attach a sole’, sen ‘eye of a needle’ > sen-mi- ‘thread a
needle’
Even -mĀ- (Benzing 1955a: 1065): ulikī ‘squirrel’ > ul-mē- ‘hunt squirrel’, tebekī
‘tobacco’ > tebekī-mē- ‘go and get tobacco’, aman ‘father’ > am-mā- ‘like the father’
Even -mī- (Benzing 1955a: 1065): ulikī ‘squirrel’ > ul-mī- ‘hunt squirrel’, olra
‘fish’ > olra-mī- ‘fish, catch fish’
Ud. -mA- (Nikolaeva 1999: 169-170): xoton ‘city’ > xoto-mo- ‘go to the city’, āda
‘mountain pass’ > āda-ma-si- ‘cross the mountain’, jege ‘reference point’ > jege-me‘orientate oneself’, kakt’a ‘half’ > kakt’a-ma- ‘divide in half’, tō ‘fire’ > to-mo‘warm on a fire’, ānta ‘woman’ > ānta-ma- ‘womanize’, ilaka ‘bast’ > ila-ma- ‘bark
lime-trees’, olondo ‘ginseng’ > olondo-mo- ‘collect ginseng’, oloxi ‘squirrel’ > oloxime- ‘hunt for squirrels’, l’asa ‘goby fish’ > l’asa-ma- ‘catch goby fish’
Benzing (1955a: 1065) and Avrorin (1961: 21) treat the denominal suffix Na. masi- ‘hunt for the base noun’ as a compound of *-ma- and *-si-. The second element is suggested to be the deverbal suffix Na. -si- for continuous, multiple or occasional actions (Avrorin 1961: 46) that also underlies in the reiterative suffix Na. nasi-, analyzed in section 4.3. above. E.g. gasa ‘goose’ > gasa-masi- ‘hunt geese’,
giu ‘deer’ > giu-mesi- ‘hunt deer’, moksa ‘hare’ > moksa-si- ‘hunt for hares’, sogda
‘fish’ > sogda-masi- ‘fish, catch fish’
How the actional suffix chain connects Japanese to Altaic
43
deadjectival
Only in Manchu was I able to trace examples of deadjectival derivation.
Ma. -mi- (Gorelova 2002: 236): bolgo ‘clean, clear’ > bolgo-mi- ‘abstain, fast’, goro
‘far’ > goro-mi- ‘do from afar, go a long distance’
deverbal
Evk. -mu- (Nedjalkov 1997: 302): ā- ‘sleep’ > ā-mu- ~ ā-me- ‘want to sleep’, bu‘die’ > bu-mu- ‘be ill’, jep- ‘eat’ > jep-mu- ‘be hungry’, um- ‘drink’ > um-mu- ‘want
to drink, be thirsty’, inekte- ‘laugh’ > ine-mu- ‘smile’
Ramstedt (1952: 181-182) and Benzing (1955a: 1069) treat the deverbal desiderative suffixes Even -mš-, Ud. -mūi-, Na. -mosi- ~ -musi- as compounds of *-muand *-si-. The second element is identified as the deverbal suffix for continuous,
multiple or occasional actions.
Ud. -m˚i- (Nikolaeva 1999: 189): umi- ‘drink’ > umi-mūi- ‘want to drink’, iñi‘laugh’ > iñi-mūi- ‘want to laugh’, ŋene- ‘go’ > ŋene-mūi- ‘want to go’, tigde- ‘rain’
> tigde-mūi- ‘be going to rain’
Na. -mosi- ~ -musi- (Avrorin 1961: 64): jep- ‘eat, feed’ > je-musi- ‘want to eat, be
hungry’, omi- ‘drink, smoke’ > omi-mosi- ‘want to drink, smoke, be thirsty’, ao‘sleep, lie down’ > a-masi- ‘want to sleep, be sleepy’
Mongolic
pMo *-mA-: WMo. -mA- ‘attempt or intend to carry out the verb base, be inclined to
carry out the verb base, be able to carry out the verb base’
pMo *-mA- + *-G deverbal noun > WMo. -mAG deverbal noun denoting inclination
(Poppe 1954: 45, 48)
pMo *-mA- + -GAi deverbal noun > WMo. -mAGAi deverbal noun denoting inclination or ability to act (Poppe 1954: 45, 48)
pMo *-mA- + -l deverbal noun > WMo. -mAl deverbal noun denoting the result of
craftsmanship (Poppe 1954: 47, 48)
pMo *-mA- + -r deverbal noun > WMo. -mAr deverbal noun denoting suitableness or
fitness (Poppe 1954: 48, 49)
The intentional suffix pMo. *-mA- has lexicalized in a number of stems. Although the semantics ‘attempt, intend, be inclined to, be about to’ became obscure in
the following examples, they are still clearly reconstructable from the composite
suffixes. Apart from WMo. –Gai, the individual suffixes forming the second part of
44
Martine Robbeets
the compound are exclusively deverbal. This suggests that *-ma- is a suffix that derives verbs.
WMo. -ma-: WMo. cir- ‘drag, pull along’ > cir-ma- ‘strive after, endeavor, make an
effort’, WMo. oi- ‘rebound, avoid, go away in another direction’ > oi-ma- ‘swim
(across), ford, fly’, WMo. kele- ‘speak, say’ (> *kele-me->) keme- ‘say, be named,
intend’,14 WMo. üi- ‘mix’ > üi-me- ‘become disturbed, bustle’
WMo. -mAG-: ide- ‘eat, consume’ > (*ide-me- ‘want to eat’) > ide-me-g ‘having a
good appetite, greedy’, egede- ‘turn sour, curdle’ > (*egede-me- ‘be about to turn
sour’) > egede-me-g ‘curdled milk, mixture of cold milk and hot buttermilk, food for
domestic animals’, jori- ‘move in the direction of, strive, be resolved’ > (jori-ma‘intend to move in a certain direction’) > jori-ma-V ‘willful, intentional, having a
purpose, courageous’
WMo. -mAGAi-: ide- ‘eat, consume’ > (*ide-me- ‘want to eat’) > ide-me-gei ‘voracious, venal’, jori- ‘move in the direction of’ > (jori-ma- ‘intend/ be able to move in
a certain direction’) > jori-ma-gai ‘enterprising, decided, resolute’, umta- ‘sleep’ >
(*umta-ma- ‘want to sleep’) > umta-m-qai ‘be sleepy’, sur- ‘learn, study, ask, inquire’ (> *sur-ma- ‘be inclined to, be able to learn’) > sur-ma-γai ‘be gifted, be
trained, be experienced’
WMo. -mAl-: jiru- ‘draw (a line, a picture)’ (> *jiru-ma- ‘be able to draw’) > jiruma-l ‘sketched, painted’, bici- ‘write, inscribe’ (> *bici-me- ‘be able to write’) > bicime-l ‘writing, manuscript, inscription’, neke- ‘knit, weave’(> *neke-me- ‘be able to
weave’) > neke-me-l ‘woven, knitted, textile’
WMo. -mAr-: ide- ‘eat, consume’ > (*ide-me- ‘want to, be possible to eat’) > ide-mer ‘edible, consumed, eaten, having a keen appetite’, üje- ‘see, behold, look at, glance
at’ (> *üje-me- ‘want to, be able/possible to see’) > üje-me-r ‘scene, sight, exhibition’
deadjectival (quality verbs)
WMo. -mAr-: γaiqa- ‘be astonished, surprised’ > γaiqa-ma-r ‘astonishing, surprising’, ayu- ‘be frightened’ > ayu-ma-r ‘horrible, frightening’
14
The derivation of WMo. keme- ‘say, be named, intend’ suggests the same liquid syllable
loss as is observed in the alternation of WMo. kele- ‘speak, say’ with WMo. kelele- ‘say,
speak’, derived from from kelen ‘word, speech, tongue’.
How the actional suffix chain connects Japanese to Altaic
45
4.6. pJ *-ka- < pA *-ga- inchoative (denominal, deadjectival, deverbal)
Ramstedt (1912: 54-59) compares the denominal and deverbal inchoatives in Mongolic and Turkic. Later (1952: 200-201) he adds the Korean denominal inchoative to
the etymology. Although he labels them ‘intensives’, Miller (1981: 867-868) compares the Japanese and Turkic deverbal suffixes. In an article (1982) devoted to this
suffix he includes Korean, Tungusic and Mongolic cognates. Poppe (1972: 123-124)
reconstructs a Tungusic transitive suffix on the basis of the derived Evenki verbs, but
in reference to the Udehe material, the original Tungusic suffix was probably an
inchoative.
The etymology discussed here is straightforward as far as the denominal inchoative is concerned. I was unable to find examples of deadjectival derivation in Tungusic. In Korean the evidence for deadjectival and deverbal derivation is rather
speculative. From a phonological perspective, the voiced velar correspondence is
regular (Robbeets 2005a: 91, 2005b). The front vowel in Mongolic is problematic.
The semantic fit is satisfactory.
Korean
pK *-k- > K, MK -k- (Ramstedt 1939: 138, Ramsey 1986, 1991, p.c.15)
‘1. enter a new state or aim at a concept similar to the base noun, 2. make use of the
base noun’
‘acquire the property denoted by the adjective’
‘start the action or process denoted by the base verb, enter the state resulting from the
verb base’
denominal
1. K mwul, MK .mul ‘water’ > K mwulk-, MK mulk-, mwulk- ‘be watery, be thin’, K
pwul, MK .pul ‘fire’ > K pwulk-, MK pulk- ‘be red, be crimson’, MK mwus ‘sheaf,
bundle’ > K mwukk-, MK mwusk- ‘tie up into a bundle’, MK kul ‘writing’ > MK
kulk- ‘scratch’, MK .pich ‘light, brilliance, color’ > MK pi∆G-/ pi∆u- ‘embellish,
decorate’ (< pK *pisuk- according to Martin 1996: 29), 2. K ōl, MK ¨ol ‘strand, ply,
warp’ > K olk- ‘tie up, bind, weave’, MK . naks ‘fishhook’ > MK naksk- ‘fish’, MK
kus ‘line, limit’ > MK ku∆G-/ku∆u- ‘draw a line, delimit’ (< pK *kusuk- according to
Martin 1996: 29)
15
I would like to express my gratitude to Professor Ramsey for sharing an unpublished list of
Middle Korean verbs that he put together a number of years ago. The verbs on the list are
sorted according to Ramsey’s form classes. When applicable, the derivational base is given
along with the verb. Although the data have been very helpful, I bear full responsibility for
their analysis.
46
Martine Robbeets
deadjectival and deverbal?
Traces of deadjectival and deverbal use of the inchoative are fragmentary and rather
obscure. The suffix *-k- may derive adjectives in e.g. pK *kis- ‘be happy’ (+ pK
*-.p(u/o)- derives depictive adjectives from adjective stems) in MK kispu- ‘be happy’
> MK kisk- ‘rejoice’, pK *polo- ‘be clear’ (+ pK *-(k)i- causative) in K poli- ‘clear
away, make clear’ > MK polk- ‘be(come) bright’,16 pK *cye- ‘be small’ (+ pK *-minclination) in MK ¨cyem- ‘be young’ > MK ¨cyek- ‘be small, few’, (pK *kul ?) >
MK ¨kulk- ‘be thick, big’. The latter two adjectives belong to accent class 5. They are
monosyllabic and have a rising tone. A rising tone usually results from the contraction of two syllables. The majority of the members of accent class 5 are derived
verbs. Apart from the accent class to which MK ¨kulk- ‘be thick, big’ belongs, I do
not find internal evidence for the reconstruction of pK *kul- ‘abundant, thick, big’.
However, there is a possible external parallel for the word in pMo *kur ~ kür ‘abundant, thick, fat’ reflected in WMo. kür, kür-tei ‘abundant’, qur-tai ‘having accumulated fat’, qur-la- ‘accumulate fat, grow thick’ (Lessing 1960: 503, 507, 991). In
Turkic we find a root pK *kür that has the meaning ‘abundant, thick, dense’ in contemporary languages such as in Tk. gür, Az. gür, Tkm. gür, Gag. gür, Karaim kür.
Tuva has xür ’healthy, well fed'. However, in Karakhanide kür is only attested as
‘stout-hearted, courageous’ (Clauson 1972: 735). Chuv. kəwrəw ‘abundant, courageous’, Bash. kör ‘well fed, courageous’ and Tat. kör ‘well fed, courageous’ preserve
both meanings. Although it is attested earlier, the meaning ‘courageous’ appears to
be a metaphorical extension of ‘abundant’ in Turkic. Whether this word is to be considered a borrowing or a cognate does not influence the argument put forward here.
Finally the evidence for the inchoative suffix in the deverbal derivation is even
fainter. Perhaps we find a trace of the suffix in verb pairs such as MK kyes- ‘experience’ > K kyekk- ‘ experience’ (< *kyesk-), MK pu∆u- ‘break’ > MK .psku- (<
*pusuk-) ‘shell, peel (tr.); hatch (intr.), MK .tum- ‘submerge’ > MK tumk- ‘immerse’,
but in the latter case we could be dealing with a reduced causative suffix -ki-.
The resultative aspect in Middle Korean is marked with the suffix MK -. ke-/-.ka
(Martin 1992: 263, 466, 595, 601; 1995). It can derive adjectives and verbs.17 The
resultative MK - .kA- is logically exclusive with the processive MK -.no- discussed in
section 4.3. Martin (1992: 261) positions the suffix to the right of the actional suffixes, but it precedes markers of politeness, mood and aspect. Apart from the seman16
17
pK *polo- ‘be clear’ would be a phonologically regular fit for four subsequent phonemes
to pJ *para- ‘be clear’ reflected in harau ‘purify, exorcize, clear', hareru ‘clear up, vanish,
be refreshed’.
An example of the suffix following an adjective is Mod. K nep-ke-na nep-un thyenha
(1747 Songkang 1:4a) (wide-result-adversative wide-modifier land beneath heaven) ‘as
wide as can be, wide land beneath heaven’. An example of the suffix following a verb is
MK ans-ke-.na sye-ke-.na (1447 Sek 19:5b) (sit-result-adversative stand-resultadversative) ‘whether (ultimately) sitting or (ultimately) standing’.
How the actional suffix chain connects Japanese to Altaic
47
tic difference between inchoative and resultative, there is a second obstacle for comparing the Korean suffix in this context. Internal evidence suggests that the resultative suffix has grammaticalized from the verb MK .ka- ‘go’. The failure of the resultative suffix MK -.ka- to occur after the verb stem MK .ka- ‘go’ suggests an unwanted iteration and a verbal origin for the resultative suffix. Although there are a
few forms with MK .wo.ke-, most of the resultative forms of MK .wo- ‘come’ become
MK .wo.na- (cfr. section 4.3). The auxiliary MK .na- ‘emerge’ probably substitutes
the auxiliary MK .ka- ‘go’ which would be incompatible with MK wo- ‘come’. Besides, many of the monosyllabic high-accent stems that end in a vowel lose the accent in the common paradigmatic forms, but they retain the accent before MK -. ke-/ .
ka (Martin 1992: 70). The peculiar accentual behavior could indicate that the resultative originally was a bound stem. These observations project the resultative suffix
MK -.ka-/ - .ke- back to the lexical level. Although the Korean suffix is not likely to
be cognate with the Altaic inchoative, as an original motional verb it could be related
to an Altaic root pA *ka- ‘go’.18
Tungusic
pTg *-gA- inchoative : Evk. -gA-, Ud. -gA‘begin to move the base noun, change the position of the base noun, bring the base
noun into existence’
‘begin (of) the action denoted by the base verb’
denominal
Evk. -gA- (Konstantinova 1964: 200, Nedjalkov 1997: 301)
iti ‘business’ > iti-ga- ‘set things going, organize, plan to do’, kolto ‘fist’ > kolto-go‘hit with the fist’, asakī ‘wing’ > asa-ga- ‘flap the wings’, saŋar ‘hole, gap’ > saŋarga- ‘pierce, make a hole, desert, break off (relations)’, usi ‘bridle, curb’ > usi-ge‘unharness a reindeer’
deverbal
Evk. -gA- (Konstantinova 1964: 161, Nedjalkov 1997: 231) inchoative transitive. It
derives transitives from intransitive verb stems with an inchoative nuance.
tati- ‘be used to (intr.)’ > tati-ga- ‘get (sb) used to, accustom (sb) to, train, learn
(tr.)’, iti- ‘come into being (intr.)’ > iti-ga- ‘set things going, organize, plan to do
18
The Korean resultative and the verb K ka-, MK .ka- ‘go’ reflect pK *ka- ‘go’. The Japanese verbs OJ ik- ‘go’, OJ kayo1p- ‘ply between, commute, frequent, go regularly’, OJ
kare2- ‘get apart, cease, go away’ and OJ kaye- ‘be parted, get apart’ reflect an original
motional root pJ *ka- ‘go (away)’. A possible Mongolic cognate is WMo. γar-, Khal. gar-,
Kalm. γar-, Dong. qeri-, Bao. xăr-, Dag. gar-, Yogh. Gar-, Mgr. Gari-, Mogh. γaru-, pMo
*gar- ‘go out’.
48
Martine Robbeets
(tr.)’, kamñi- ‘defend oneself (intr.)’ > kamñi-ga- ‘attack, tackle (tr.)’, kapu- ‘burst,
explode, break, crack (intr.)’ > kapu-ga- ‘break, tear in two (tr.)’, Ev. ada- ‘fly past
(intr.)’ > ada-ga- ‘try to escape (tr.)’, bele- ‘help (intr.)’ > bele-ge- ‘help (tr.)’
Ud. -gA- (Nikolaeva 1999: 178) inchoative decausative. It derives the decausative
counterpart of the transitive verbs with the inchoative suffix -li-. It expresses the
beginning of a state caused by a transitive base verb.
guza-li- ‘tear’ > guza-ga- ‘get torn’, bukta-li- ‘break in two’ > bukta-ga- ‘get broken in two’, xudu-li- ‘dislocate’ > xudu-ge- ‘get dislocated’
Nanai has a denominal inchoative -go- ~ -gu- that is particularly used for the beginning of a cyclic natural phenomenon (Menges 1968: 202-203; Avrorin 1961: 2021, 56), e.g. Na. joa ‘summer’ > joa-go- ‘come (of summer)’, Na. dolbo ‘night’ >
dolbo-go- ‘come (of night)’, Na. ñeŋñe ‘spring’ > ñeŋñe-gu- ‘come (of spring)’, sikse
‘evening’ > sikse-gu- ‘come (of evening)’, tue ‘winter’ > tue-gu- ‘come (of winter)’.
The same suffix occurs following ordinal adjectives such as Na. juyecie ‘second’ >
juyecie-gu- ‘do for the second time’, Na. iliacia ‘third’ > iliaci-gu- ‘do for the third
time’, Na. duyecie ‘fourth’ > duyecie-gu- ‘do for the fourth time’. Following verbs it
has a repetitive or regressive meaning e.g. Na. bu- ‘give’ > bu-gu- ‘give back’, Na.
uŋ- ‘say, talk’ > uŋ-gu- ‘say again, repeat’, Na. ili- ‘stand up’ > ili-go- ‘stand up
again, stand up after sitting’. In Udehe the suffix has a reflex -gi-. It is used as a denominal inchoative following nouns, e.g. Ud. bolo ‘fall’ > bolo-gi- ‘come (of fall)’,
Ud. neki ‘spring’ > neki-gi- ‘come (of spring) , Ud. dogbo ‘night’ > dogbo-gi- ‘come
(of night)’, Ud. tue ‘winter’ > tue-gi- ‘come (of winter)’. It also derives repetitives or
regressives from verbs, e.g. Ud. deu- ‘get tired’ > deu-gi- ‘get tired again’, Ud. nodo‘lose’ > nodo-gi- ‘lose again’, Ud. jawa- ‘take’ > jawa-gi- ‘take back’, Ud. tama‘pay’ > Ud. tama-gi- ‘pay debts’, Ud. pou-pou ‘dark’ > pou-te-gi- ‘get dark’, Ud.
bagdi- ‘live’ > bagdi-gi- ‘be born’. Since the common denominator appears to be
‘repetitive’ rather than ‘inchoative’, I am reluctant to include this suffix in the comparison.
Mongolic
pMo *-gi- > WMo. -gi- ~ -ki- (sporadic fortition after consonants) / -i- (lenition after
vowels)
‘enter or bring into a new state similar to the base noun’
‘acquire the property denoted by the adjective’
‘1. start the action or process denoted by the base verb, 2. enter the state resulting
from the verb base’
How the actional suffix chain connects Japanese to Altaic
49
denominal
usun ‘water’ > usubki- ‘be watery, liquid, tasteless’, mösün ‘strand of rope’ > musgi~ muski- ‘to twist, strand (rope)’, culcu-i- ‘become inflated, blown up’ (compare
Evk. cilcun, Ma. cilcin, Na. cilcu-kte, Olč. culcu-kte ‘swelling, gland’, whether
borrowing or cognate to the Mongolic base noun), gede ‘nape or back of the neck,
occiput’ > gede-i- ‘throw back one’s head, bend backwards’.
deadjectival
ca-gan ‘white, light (of color)’, ca-kir ‘very white, snow white, white spots on finger
nails or on the feathers of a bird’> ca-i- ‘become white, turn pale, turn grey, to dawn,
grow light’, ula ‘red (adj.)’ > ula-i- ‘become red, redden, blush’, dere-gir ‘stiff and
sticking out’ > dere-i- ‘become stiff’, soqur ‘blind (adj.)’, soqu-la- ‘make blind (tr.)’
> soqu-i- ‘be(come) blind (used as an invective)’, taki-r ‘crippled, crooked, a cripple’
> taki-i- ‘become crippled, bent, oblique (intr.)’
deverbal
1. ala- ‘kill, murder, butcher (tr.)’ > al-ki- ‘hit, beat (tr.)’, kel- ‘to be strung (as
pearls), to follow in succession (intr.)’ > kel-ki- ‘to string (pearls), to thread, to bring
together (tr.)’, sedü- ‘think out, work out, invent (tr.)’ > sedki- ‘think, reflect, intend
(intr. / tr.)’, alab-ca- ‘kick or strike with the foot’ > alab-ki- ‘jump onto a horse’, jilu‘flee, run away, avoid (intr.)’ > jili-i- ‘flee, take flight (intr.)’, 2. er-ci- ‘twist, spin
(thread or rope) (tr.)’, er-cim ‘torsion, twist, steadfastness’ > er-gi- ‘turn or move
around, revolve, circumambulate (intr./ tr.)’, tala- ‘take away, confiscate, plunder,
ruin (tr.)’ > tali-i- ‘disappear, get lost, go astray, die (intr.)’ (L 773), jimu- ‘sink, go
down (intr.)’ > jimi-i- ‘become tightened (of lips, eyes) (intr.)’.
Turkic
pTk *-k- > OTk. -(X)k- inchoative (Gabain 1950: 82, Erdal 1991: 492-499, 524, 645649)
‘enter or bring into a new state similar to the base noun, begin (of a cyclic natural
phenomenon)’
‘acquire the property denoted by the adjective’
‘1. start the action or process of the intransitive verb base, 2. come into the state resulting from the transitive verb base’
denominal
ada ‘danger’ > ada-k- ‘to be or come into distress’, ant ‘oath’ > ant-ïk- ‘swear an
oath’, bir ‘one’ > bir-ik- ‘get together, join (tr. / intr.), become united’, šï ~ čï ‘moist’
> čï-k- ‘to get moist’, ät ‘flesh, meat’ > ät-ik- ‘put on flesh’, tag ‘mountain’ > tag-ïk‘to go to the mountains’, uya ‘animal’s resting place, bird’s nest’ > uya-k- ‘set (of
50
Martine Robbeets
sun, stars)’, yer ‘place, land’ > yer-ik- ‘settle’, yay ‘spring’ > yay-ïk- ‘become
spring’, küz ‘autumn’ > küz-ük- ‘turn to autumn, become autumnal’
deadjectival (quality verbs)
*alï- ‘be bad’ in alïg ‘bad’ > alï-k- ‘turn septic, fester, detoriate’ (EDT 138), *amrï‘be peaceful’ in amrïl- ‘be at peace, be at ease’ > amrï-k- ‘acquire peace, become
peaceful’, us- ‘be thirsty’ > us-uk- ‘be thirsty, be overcome with thirst’, ač- ‘be hungry’ > ač-uk- ‘be famished, be overcome with hunger’
deverbal
1. čom- ‘sink (intr.)’ > čom-uk- ‘drown (intr.)’, tal- ‘lose strength, faint, be sunk (in
thought, sleep) (intr.)’ > tal-ïk- ‘go under (intr.)’, oŋ- ‘turn pale, fade, wilt (intr.)’ >
oŋ-uk- ‘become pale, become lean because of illness, wilt’, öč- ‘go out, be extinguished (of fire) (intr.)’ > öč-ük- ‘fail (of voice), be interrupted (of breathing), die (of
embers) (intr.)’, 2. sor- ‘ask, inquire about (tr.)’ > sor-uk- ‘be inquired about (intr.)’,
tar- ‘disperse, scatter, do away with (tr.)’ > tar-ïk- ‘disperse, be driven away, go
away (intr.)’
5. Conclusion
The question whether Japanese is related to the Altaic languages has been treated in
the literature for nearly two centuries. Lexical comparison yields positive evidence,
but comparative morphology remains relatively unexplored. The present paper examines verbal morphology. The focus is on derivational and actional suffixes relating
Japanese to Altaic.
The choice of the topic is motivated by Johanson’s (2002) observation that the
positions close to the primary stem are particularly diagnostic for genetic continuity.
This is an important point of departure that is overlooked in Vovin’s (2001) study of
verbal morphology. His article is mainly oriented on the markers in the outer ranks
for aspect, mood, tense, final predication. Many of Vovin’s etymologies are in contradiction with the internal analysis of the compared form. They often disregard
processes of grammaticalization within Japanese (Robbeets 2005a: 157-173). Discouraged by the weakness of the evidence, Vovin today (2005) rejects the Altaic
hypothesis.
The scope of this study is broad in the sense that the comparisons stretch over
five branches: Japanese, Korean, Tungusic, Mongolic and Turkic. This approach
differs from morphological studies that do not include Japanese such as Ramstedt
1912, 1952, Poppe 1972, Baskakov 1981, Kormušin 1984, or that exclude Turkic and
Mongolic such as Vovin (2001), or that compare Japanese and Korean from a binary
perspective such as Whitman (1985) and Martin (1968, 1990, 1991a, 1995). Particularly when dealing with morphology, a global approach is more promising because
How the actional suffix chain connects Japanese to Altaic
51
traces of conservative items are expected in geographically and linguistically remote
areas such as Japanese and Turkic.
A premise that underlies this study is that in all the Altaic languages verbs, adjectives and nouns are distinct parts of speech. This assumption goes against the viewpoint taken in Miller (1982) and Menges (1975, 1978). It has consequences for the
reconstruction and comparison of the suffixes under investigation. The theoretical
base for the distinction is taken from studies such as Dixon (1982) and Johanson
(2006).
The argumentation in this article is data-oriented. Recent years have seen a sudden improvement of our access to the data. The present work relies on the comprehensive reconstruction of the proto-Japanese verbal system advanced in Unger 1977
and Martin 1987 and on an analysis of Korean verbs by Ramsey (1986, 1991) and
Martin (1992). It further benefits from recent progress in the description of individual
Tungusic languages made by Malchukov (1995), Nedjalkov (1997), Nikolaeva
(1999) and Gorelova (2002) and from up-to-date contributions to the analysis of
Middle and Old Turkic made by Berta (1996) and Erdal (1991, 2004). As far as the
evaluation of the phonological correspondences is concerned, I build on the findings
in my previous work (Robbeets 2005a).
The conclusions can be summarized in the following table. I use the abbreviations
N for denominal, O for onomatopoetic, A for deadjectival and V for deverbal.
pJ
*-raeffort
N-O-A
pK
*-yatransformation
N-A
pTg
*-lĀeffort
N-A
pMo
*-lAeffort
N-O-A
pTk
*-lAeffort
N-O-A
pA
*-laeffort
N-O-A
*-dĀtransform.
N-A
*-d(A)transform.
N-A
* -(A)dtransform.
N-A
*-datransform.
N-A
*-naprocess
N-A-V
*-noprocess
A-V
*-nA*-n(A/i)natural process spontaneity
N-A-V
A-V
*-nspontaneity
A-V
*-naprocess
A-V
*-kaiconic
O
*-kiiconic
O
*-kiiconic
O
*-kiiconic
O
-kIiconic
O
*-kiiconic
O
*-mainclination
N-O-A-V
*-m(u/o)inclination
N-A(-V?)
* -m(a/u/o)intention
N-A (-V?)
*-mAintention/ability intention
A-V
*-ma-
*-kainchoative
N-A-V
*-kinchoative
N-A (-V?)
*-gAinchoative
N-V
*-giinchoative
N-A-V
*-gainchoative
N-A-V
*-kinchoative
N-A-V
N-A-V
52
Martine Robbeets
The actional suffix chain reconstructed for Japanese in section 3.1. can be etymologized for every single suffix. The etymologies stretch over at least four branches. The
cognates are global in the sense that the correspondences are material, semantic,
derivational and combinational. The consonants correspond without exception in
accordance with the system established on the basis of the lexical material in Robbeets 2005a. The vowel correspondences are satisfactory because they show few
irregularities. The lack of the final vowel in most of the Turkic reflexes is probably
connected with the gradual loss of final reduced vowels in Old Turkic (Johanson
1979). The semantic latitude is very limited. The parts of speech to which the derivational bases belong are parallel in a way that a reconstruction of the derivational
behavior in Altaic is possible. The order of the suffixes in the table is the relative
suffix order as it occurs in the Japanese chain. It roughly overlaps with the distributional characteristics of the suffixes in the other languages.
The regularity and the systematics of the shared properties summarized in the table, taken together with the fact that they confirm the phonological correspondences
established in earlier work, exclude coincidence as a possible motivation of the
similarities. The etymologies discussed in this article are not the result of sheer
chance. Neither can they be explained by universal tendencies in linguistic structuring. The remaining possibilities are either borrowing or common ancestorship. In my
opinion it is more difficult to attribute the similarities to borrowing than to attribute
them to common ancestorship. The correspondences are global, regular and symmetrical. The etymologies are spread over five branches, which makes it hard to explain
copies all the way from Turkic into Japanese. The evidence consists of bound, verbal
morphemes in a position close to the primary stem. With all due caution, I cannot but
attribute the etymologies presented here to the common ancestorship of the languages under comparison. The actional suffix chain most probably connects Japanese
to Altaic in a genealogical sense.
References
Avrorin, Valentin A. 1961. Grammatika Nanajskogo jazyka. Tom 2. Morfologija glagol’nych i
narečnych častej reči, meždometij, služevnych slov i castišč. Moskova: Nauk.
Backhouse, Anthony E. 1984. Have all the adjectives gone? Lingua 62,169-86.
Baskakov, Nikolaj A. 1981. Altaiskaja sem’ja jazykov i ee izučenie. Moscow: Nauk.
Benzing, Johannes 1955a. Die tungusischen Sprachen. Versuch einer vergleichenden Grammatik. Abhandlungen der geistes- und sozialwissenschaftlichen Klasse 11, 949-1099.
Benzing, Johannes 1955b. Lamutische Grammatik, mit Bibliographie, Sprachproben und
Glossar. Wiesbaden: Otto Harrassowitz.
Berta, Árpád 1996. Deverbale Wortbildung im Mittelkiptschakisch-Türkischen. (Turcologica
24.) Wiesbaden: Harrassowitz.
Cincius, Vera I. (ed.) 1975-77. Sravnitel’nyj slovar’ tunguso-man’čžurskich jazykov. 2 vols.
Moskva: Nauka.
How the actional suffix chain connects Japanese to Altaic
53
Clauson, Gerard 1972. An etymological dictionary of pre-thirtheenth-century Turkish. Oxford:
Clarendon Press.
Comrie, Bernard 1976. Aspect: an introduction to the study of verbal aspect and related
problems. Cambridge: University Press.
Dixon, Robert M. W. 1982. Where have all the adjectives gone? In: Dixon, Robert M.W. 1982
(ed.) 1982. Where have all the adjectives gone? Amsterdam: Mouton. 1-62.
Doerfer, Gerhard 1963-75. Türkische und mongolische Elemente im Neupersischen, unter
besonderer Berücksichtigung älterer neupersischer Geschichtsquellen, vor allem der
Mongolen- und Timuridenzeit. 4 vols. Wiesbaden: Franz Steiner.
Doerfer, Gerhard 1974. Ist das Japanische mit den altaischen Sprachen verwandt? Zeitschrift
der Deutschen Morgenländischen Gesellschaft 124, 103-142.
Doerfer, Gerhard & Hesche, Wolfram & Scheinhardt, Hartwig 1980. Lamutisches Wörterbuch.
Wiesbaden: Harrassowitz.
Erdal, Marcel 1991. Old Turkic word formation. A functional approach to the lexicon. (Turcologica 7.) Wiesbaden: Harrassowitz.
Erdal, Marcel 2004. A grammar of Old Turkic. (Handbook of Oriental Studies) Leiden: Brill.
Finch, Roger 1985. Particles used with the ‘absolutive case’ in the Altaic Languages. Journal
of Turkish Studies 9, 27-36.
Finch, Roger 1987. Verb classes in the Altaic languages. Sophia Linguistica 26, 41-61.
Finch, Roger 1999. The case system of the Altaic languages. Surugadai University Studies 18,
87-112.
Gabain, Annemarie von 1950. Alttürkische Grammatik. Leipzig: Otto Harrassowitz.
Gorelova, Liliya M. 2002. Manchu grammar. (Handbook of Oriental Studies) Leiden: Brill.
Grønbech, Karl 1936. Der türkische Sprachbau. Kopenhagen: Levin & Munksgaard.
Haenisch, Erich 1939. Wörterbuch zu Manghol un Niuca Tobca’an. Leipzig: Harrassowitz.
Haenisch, Erich 1961. Mandschu Grammatik. Leipzig: Otto Harrassowitz.
Hangin, Gombojab 1986. A Modern Mongolian-English dictionary. Bloomington: Research
Institute for Inner Asian Studies.
Harrison, Sheldon P. 2003. On the limits of the comparative method. In: Joseph,
Brian D. & Janda, Richard D. (eds.) 2003. The handbook of historical linguistics.
Oxford: Blackwell. 213-43.
Hattori, Shirō 1959. Nihongo no keitō. Tokyo: Iwanami shoten.
Itabashi, Yoshizō 1988. A comparative study of the Old Japanese accusative case suffix wo
with the altaic accusative case suffixes. Central Asiatic Journal 32, 193-231.
Itabashi, Yoshizō 1989. The origin of the Old Japanese prosecutive case suffix yuri. Central
Asiatic Journal 33, 47-66.
Itabashi, Yoshizō 1990. The origin of the Old Japanese accusative case suffix i. Ural-Altaische
Jahrbücher Neue Folge 9, 152-173.
Itabashi, Yoshizō 1991. The origin of the Old Japanese genitive case suffixes *n / nö / na / Nga
and the Old Korean genitive case suffix *i in comparison with Manchu-Tungus, Mongolian, and Old Turkic. Central Asiatic Journal 35, 231-78.
54
Martine Robbeets
Itabashi, Yoshizō 2001. Arutaishogengo (ganchōsengo) to nihongo. Gengo Kenkyū 120, 106116.
Janhunen, Juha 1992. Das Japanische in vergleichender Sicht. Journal de la Société finnoougrienne 84, 145-161.
Janhunen, Juha 1994. Additional notes on Japanese and Altaic. Journal de la Société finnoougrienne 85, 236-40.
Janhunen, Juha 2003. Proto-Mongolic. In: Janhunen, Juha (ed.) 2003. The Mongolic languages. London: Routledge. 1-29.
Johanson, Lars 1975. Aktionalphrase und Verlaufsordnung. In margine einer Monographie zur
„Aktionalität” im Deutschen. Studia Neophilologica 47, 120-150.
Johanson, Lars 1977. Zur Theorie der ‚Aktionalität’. Studia Neophilologica 49, 119-128.
Johanson, Lars 1979. Alttürkisch als ‘dissimilierende Sprache’. (Abhandlungen der Akademie
der Wissenschaften und der Literatur, Mainz, Geistes- und sozialwissenschaftliche Klasse
3.) Wiesbaden: Steiner.
Johanson, Lars 1992. Strukturelle Faktoren in türkischen Sprachkontakten. (Sitzungsberichte
der Wissenschaftlichen Gesellschaft an der J.W.Goethe-Universität Frankfurt am Main.
29, 5.) Stuttgart: Steiner.
Johanson, Lars 1999. Cognates and copies in Altaic verb derivation. In: Menges, Karl Heinrich
& Neumann, Nelly (eds.) 1999. Language and literature: Japanese and the other Altaic
languages: Studies in honour of Roy Andrew Miller on his 75th birthday. Wiesbaden: Harrassowitz. 1-14.
Johanson, Lars 2002. Structural factors in Turkic language contacts. Richmond: Curzon.
Johanson, Lars. 2006. Adjectives and nouns in South Siberian and other Turkic languages. In:
Erdal, Marcel & Nevskaya, Irina 2006. Exploring the Eastern frontiers of Turkic. (Turcologica 60.) Wiesbaden: Harrassowitz. 57-78
Kaiser, Stefan; Ichikawa, Yasuko & Kobayashi, Noriko 2001. Japanese: A Comprehensive
Grammar. London: Routledge.
Kiyose, Gisaburō Norikura 2002. The relationship among the languages of the Korean Three
Kingdoms, Proto-Japanese and Tungusic, based on historical sources. Transactions of the
47th International Conference on Eastern Studies. 121-122.
Kolesnikova V. D. 1966. Sintaksis evenkijskogo jazyka. Moscow: Nauk.
Konstantinova, Ol’ga A. 1964. Evenkijskij jazyk. Moscow: Nauk.
Kormušin, Igor V. 1984. Sistemy vremen glagola v altajskich jazykach. Moscow: Nauka.
Kortlandt, Frederik 1997. Japanese aru, iru, oru ‘to be’. Studia Etymologica Cracoviensia 2.
167-170.
Lee, Iksop & Ramsey, Samuel Robert 2000. The Korean language. New York: State University of New York Press.
Lee, Ki-Mun. 1977. Geschichte der Koreanischen Sprache. Wiesbaden: Dr. Ludwig Reichert
Verlag.
Lessing, Ferdinand D. (ed.) 1960. Mongolian-English dictionary. Los Angeles: University of
California Press.
How the actional suffix chain connects Japanese to Altaic
55
Lewin, Bruno 1989. VI Herkunft. In: Lewin, Bruno (ed.) 1989. Sprache und Schrift Japans.
Leiden: Brill, 98-118.
Lyovin, Anatole 1997. An introduction to the languages of the world. Oxford: University
Press.
Malchukov, Andrej L. 1995. Even. (Languages of the world. Materials 12.) München:
LINCOM.
Martin, Samuel Elmo 1968. Grammatical elements relating Korean to Japanese. In: Current
Anthropology 8, 405-407.
Martin, Samuel Elmo 1987. The Japanese language through time. New Haven: Yale University Press.
Martin, Samuel Elmo 1990. Morphological clues to the relationships of Japanese and Korean.
In: Baldi, Philip (ed.) 1990. Linguistic change and reconstruction methodology. (Trends in
Linguistics. Studies and Monographs 45.) Berlin: Mouton de Gruyter. 483- 509.
Martin, Samuel Elmo 1991a. Recent research on the relationships of Japanese and Korean. In:
Lamb, Sydney M. & Mitchell E. Douglas (eds.) 1991. Sprung from some common source.
Investigations into the prehistory of languages. Stanford: University Press, 269-292.
Martin, Samuel Elmo 1991b. A reference grammar of Japanese. Tokyo: Tuttle.
Martin, Samuel Elmo 1992. A reference grammar of Korean. Tokyo: Tuttle.
Martin, Samuel Elmo. 1995. On the prehistory of the Korean grammar: Verb Forms. Korean
Studies 19, 139-150.
Martin, Samuel Elmo 1996. Consonant lenition in Korean and the macro-Altaic question.
Honolulu: University of Hawaii Press.
Martin, Samuel Elmo 1997. Un-Altaic features of the Korean verb. Journal of Korean Linguistics 6, 3-40.
Martin, Samuel Elmo & Lee Yang-Ha & Chang Sung-Un 1967. A Korean-English dictionary.
New Haven: Yale University Press.
Menges, Karl Heinrich 1960. Morphologische Probleme. I. Zum Genitiv und Accusativ. (Uralaltaische Bibliothek 9.) Wiesbaden: Harrassowitz.
Menges, Karl Heinrich 1968. Die tungusischen Sprachen. (Handbuch der Orientalistik 1. Der
Nahe und der Mittlere Osten 5. Altaistik 3. Tungusologie.) Leiden: Brill.
Menges, Karl Heinrich 1975. Altajische Studien, II. Japanisch und Altajisch. (Abhandlungen
für die Kunde des Morgenlandes. 41.3.) Wiesbaden: Steiner
Menges, Karl Heinrich 1978. Problems of Tungus linguistics. Anthropos 73, 382-40
Menges, Karl Heinrich 1984. Korean and Altaic. A preliminary sketch. Central Asiatic Journal 28, 234-295.
Miller, Roy Andrew 1971. Japanese and the other Altaic languages. Chicago: The University
of Chicago Press.
Miller, Roy Andrew 1981. Altaic origins of the Japanese verb classes. In: Arbeitman, Yoël L.
(ed.) 1981. Bono homini donum. Essays in historical linguistics in memory of J. Alexander
Kerns. Amsterdam: Benjamins, 815-880.
56
Martine Robbeets
Miller, Roy Andrew 1982. Japanese evidence for some Altaic denominal verb-stem derivational suffixes. Acta Orientalia Academiae Scientarum Hungaricae 36, 391-403.
Miller, Roy Andrew 1985. Altaic connections of the Old Japanese negatives. Central Asiatic
Journal 31, 35-84.
Miller, Roy Andrew 1993. On some petrified case formations in the Altaic languages. Acta
Orientalia Academiae Scientarum Hungaricae 46, 288-310.
Miller, Roy Andrew & Street, John 1975. Altaic elements in Old Japanese. Part 1. Madison:
The authors. [Manuscript]
Murayama Shichirō 1957. Vergleichende Betrachtung der Kasus-Suffixe im Altjapanischen.
In: Pritsak, Omeljan (ed.) 1957. Studia Altaica. Festschrift für Nikolaus Poppe zum 60.
Geburtstag am 8 August 1957. Wiesbaden: Harrassowitz. 126-31.
Murayama, Shichirō 1975. Altaische Komponente der japanischen Sprache. In: Róna-Tas,
András (ed.) 1975. Researches in Altaic languages: Papers read at the 14th meeting of the
Permanent International Altaistic Conference held in Szeged, August 22-28, 1971. Budapest: Kiadó. 181-88.
Nedjalkov, Igor V. 1997. Evenki. Descriptive grammar. London: Routledge.
Nichols, Johanna 1992. Linguistic diversity in space and time. Chicago: University of Chicago
Press.
Nikolaeva, Irina Alekseevna 1999. A grammar of Udehe. [Doctoral dissertation, Leiden University]
Norman, Jerry 1978. A concise Manchu-English lexicon. Seattle: University of Washington
Press.
Onenko, S.N. 1980. Nanajsko-russkij slovar’. Moskva: Izdatel’stvo ‘Russkij Jazyk’.
Omodaka, Hisakata et al. (eds.) 1967. Jidaibetsu kokugo daijiten, jōdaihen. Tokyo: Sanseidō.
Petrova, Taisiia I. 1960. Nanaisko-russkij slovar’. Leningrad: Nauka.
Poppe, Nicholas 1954. Grammar of Written Mongolian. Wiesbaden: Harrassowitz.
Poppe, Nicholas 1972. Über einige Verbalstammbildungssuffixe in den altaischen Sprachen.
Orientalia Suecana 21, 119-141.
Ramer, Alexis Manaster & Vovin, Alexander & Sidwell, Paul 1997. On body part terms as
evidence in favor of the Altaic hypothesis. Ural-Altaische Jahrbücher Neue Folge 15, 116138.
Ramsey, Samuel Robert 1986. The inflecting stems of Proto-Korean. Language Research 22,
183-194.
Ramsey, Samuel Robert 1991. Proto-Korean and the origin of Korean accent. In: Boltz, William G. & Shapiro, Michael C. (eds.) 1991. Studies in the historical phonology of Asian
languages. (Current Issues in Linguistic Theory 77.) Philadelphia: Benjamins. 215-38.
Ramstedt, Gustaf John 1912. Zur Verbalstammbildungslehre der mongolisch-türkischen
Sprachen. Journal de la Société finno-ougrienne 28, 1-86.
Ramstedt, Gustaf John 1924. A comparison of the Altaic languages with Japanese. Transactions of the Asiatic Society of Japan. 2nd ser. 1, 41-54.
How the actional suffix chain connects Japanese to Altaic
57
Ramstedt, Gustaf John 1939. A Korean grammar. (Mémoires de la Société finno-ougrienne
82,1.) Helsinki: Suomalai-Ugrilainen Seura.
Ramstedt, Gustaf John 1951. Über onomatopoetische Wörter in den altaischen Sprachen.
Journal de la Société finno-ougrienne 55.2.
Ramstedt, Gustaf John 1952. Einführung in die altaische Sprachwissenschaft: II, Formenlehre. (Mémoires de la Société finno-ougrienne 104, 2.) Helsinki: Suomalai-Ugrilainen
Seura.
Robbeets, Martine Irma 2005a. Is Japanese related to Korean, Tungusic, Mongolic and
Turkic? (Turcologica 64.) Wiesbaden: Harrassowitz.
Robbeets, Martine Irma 2005b. Does Doerfer’s Zufall mean ‘cognate’? The case of the initial
velar correspondence in Altaic. Turkic Languages 8, 146-178.
Róna-Tas András. 1998. The reconstruction of Proto-Turkic and the genetic question. In:
Johanson, Lars & Csató, Éva Ágnes (eds.). 1998. The Turkic languages. New York:
Routledge. 67-80.
Rozycki, William 1994. Mongol elements in Manchu. (Uralic and Altaic Series 157.) Indiana:
Bloomington.
Schönig, Claus 2003. Turko-Mongolic Relations. In: Janhunen, Juha (ed.) 2003. The Mongolic
languages. London: Routledge. 403-419.
Shibatani, Masayoshi 1990. The languages of Japan. Cambridge: University Press.
Shōgaito, Masahiro 2002. Itabashi Yoshizō "Arutaishogengo (ganchōsengo) to nihongo." ni
taisuru komento. Gengo Kenkyū 121, 112-121.
Sohn, Ho-Min 1999. The Korean language. Cambridge: Cambridge University Press.
Starostin, Sergej Anatolevič 1991. Altajskaja problema i proisxoždenie japanskogo jazyka.
Moskva: Nauka.
Starostin Sergei & Dybo, Anna & Mudrak, Oleg 2003. Etymological dictionary of the Altaic
languages. (Handbook of Oriental Studies.) Leiden: Brill.
Steinitz, Renate 1981. Der Status der Kategorie „Aktionsart” in der Grammatik (oder: Gibt es
Aktionarten im Deutschen?) (Linguistische Studien A 76.) Berlin: Akademie der Wissenschaften der DDR.
Street, John 1957. The language of the Secret History of the Mongols. New Haven: American
Oriental Society.
Street, John 1977. Altaic elements in Old Japanese. Part 2. Madison: the Author. [Manuscript]
Sunik, Orest Petrovič. 1962. Glagol v tunguso-man’čžurskich jazykach. Moskva: Nauka.
Takeuchi, Lone 1999. The Structure and History of Japanese. New York: Longman.
Tekin, Talât 1968. A grammar of Orkhon Turkic. (Indiana University Publications. Uralic and
Altaic Series 69.) The Hague: Mouton.
Tekin, Talât 1982. On the structure of Altaic echoic verbs in {-KIrA}. Acta Orientalia Academiae Scientarum Hungaricae 36, 503-13.
Thorpe, Maner L. 1983. Ryūkyūan language history. [Doctoral dissertation, University of
Southern California, Los Angeles]
58
Martine Robbeets
Trask, Robert Lawrence. 2000. The dictionary of historical and comparative linguistics. Edinburgh: University Press.
Unger, James Marshall 1977. Studies in Early Japanese morphophonemics. Indiana: University Linguistics Club. [Doctoral dissertation, Yale University]
Unger, James Marshall & Tomita, Yoko Ito 1983. The classification of Old Japanese adjectives. Papers in East Asian Languages 1, 52-65.
Unger, James Marshall 1990. Japanese and what other Altaic languages? In: Baldi, Philip (ed.)
1990. Linguistic change and reconstruction methodology. (Trends in Linguistics. Studies
and Monographs 45.) Berlin: Mouton de Gruyter. 547-561.
Unger, James Marshall 2000. Reconciling comparative and internal reconstruction: the case of
Old Japanese /ti, ri, ni/. Language 76, 655-681.
Vasilevič, Glafira M. 1940. Očerk grammatiki evenkijskogo (tungusskogo) jazyka. Leningrad:
Učpedgiz.
Vasilevič, Glafira M. 1958. Evenkijsko-russkij slovar’. Moskva: Izdatel’stvo Inostrannych i
Nacional’nych Slovarej.
Vovin, Alexander 1994. Long-distance relationships, reconstruction methodology, and the
origins of Japanese. Diachronica 11, 95-114.
Vovin, Alexander 1998. Altaic so far? Migracijske Teme 15, 155-213.
Vovin, Alexander 2001. Japanese, Korean and Tungusic. Evidence for genetic relationship
from verbal morphology. In: David B. Honey and David C. Wright (eds.) 2001. Altaic affinities : Proceedings of the 40th meeting of the PIAC, Provo, Utah 1997. Indiana University: Research Institute for Inner Asian Studies. 183-202.
Vovin, Alexander 2003. A reference grammar of Classical Japanese prose. Richmond: Curzon.
Vovin, Alexander 2005. The end of the Altaic controversy. In memory of Gerhard Doerfer.
Central Asiatic Journal 49, 71-131.
Wang, William S-Y. & Ogura, Mieko 1999. Explorations in the origins of the Japanese language. In: Omoto, Keiichi (ed.) 1999. Interdisciplinary perspectives on the origins of the
Japanese. Tokyo: International Research Center for Japanese Studies. 309-334.
Whitman, John Bradford 1985. The phonological basis for the comparison of Japanese and
Korean. [doctoral dissertation, Harvard University, Cambrige]
Yu Changton 1973. Yi-co kwuk.e sa yenkwu. Seoul: Senmyeng munhwa-sa.
How the actional suffix chain connects Japanese to Altaic
59
Abbreviations
Az.
Bao.
Bash.
Chuv.
Dag.
Dong.
Evk.
J
Jur.
K
Kalm.
Khal.
Ma.
MJ
MK
MMo.
Mgr.
Mo.
Mogh.
Azerbaijani
Bao’an
Bashkir
Chuvash
Dagur
Dongxiang (Santa)
Evenki (Tungus, Ch. Elunchun)
(standard Tokyo) Japanese
Jurchen
(standard Seoul) Korean
Kalmuk
Khalkha
Manchu
Middle Japanese
Middle Korean
Middle Mongolian
Monguor
Mongolian
Moghol
MTk.
Na.
Neg.
OJ
Olč.
OTk.
pA
pJ
pK
pMo
pTg
pTk
SH
Tat.
Tk.
Tkm.
Ud.
WMo.
Middle Turkic
Nanai (Goldi, Ch. Hezhe)
Negidal
Old Japanese
Olcha (Ulcha, Ulchi, Olchi)
Old Turkic
proto-Altaic
proto-Japanese
proto-Korean
proto-Mongolic
proto-Tungusic
proto-Turkic
Secret History of the Mongols
(Volga) Tatar
Turkish
Turkmen
Udehe (Ude, Udege)
written Mongolian
Acknowledgements
This article is written with the support of the Humboldt Foundation, creating the
facilities for my research fellowship at the Seminar für Orientkunde of the Johannes
Gutenberg University of Mainz in 2006-2007.